Topic Chains for Understanding a News Corpus
Dongwoo Kim and Alice Oh
KAIST
Computer Science Department
Daejeon, Korea
dw.kim@kaist.ac.kr, alice.oh@kaist.edu
Abstract. The Web is a great resource and archive of news articles for
the world. We present a framework, based on probabilistic topic modeling, for uncovering the meaningful structure and trends of important
topics and issues hidden within the news archives on the Web. Central
in the framework is a topic chain, a temporal organization of similar topics. We experimented with various topic similarity metrics and present
our insights on how best to construct topic chains. We discuss how to
interpret the topic chains to understand the news corpus by looking at
long-term topics, temporary issues, and shifts of focus in the topic chains.
We applied our framework to nine months of Korean Web news corpus
and present our findings.
1
Introduction
The Web is a convenient and enormous source for learning about what is happening in the world. One can go to the Web site of any major news outlet or a
portal site to get a quick overview of the important issues of the moment. However, it is difficult to use the Web to understand what has been happening over
an extended period of time. We propose a computational framework based on
probabilistic topic modeling to analyze a corpus of online news articles to produce results that show how the topics and issues emerge, evolve, and disappear
within the corpus.
The problem of understanding a corpus of news articles over an extended
period of time is challenging because one has to discover an unknown set of topics
and issues from a large corpus of disparate sources, find and cluster similar topics,
discover any short-term issues, and identify and display how the topics change
over time. A narrower but similar problem has been studied in the TDT (topic
detection and tracking) field [1] where the goal is to identify new events and track
how they change over time. The events, however, are defined as happenings at
certain places at certain times, and so they compose a small subset of general
news topics and issues. For example, an earthquake in Haiti is an event, but the
prolonged decline of real estate sales is not. The latter makes up a large portion of
news, but the TDT community would only cover the former, whereas our research
covers both. The probabilistic topic modeling community offers solutions such
as Dynamic Topic Models [2] and Topics Over Time [3] for discovering topics
and looking at how they change over time, but those models do not capture how
topics newly emerge and disappear because they assume the same set of topics
exist from the beginning through the end of the time-series data.
We propose topic chains, a framework for analyzing a sequential corpus,
composed of similar topics appearing within a specified sliding window. Topic
chains present a temporal and similarity-based organization of topics found by
latent dirichlet allocation (LDA) [4]. Topic chains can be used to identify general
topics, such as labor unions or the stock market, which occur in long topic chains.
Short-term issues, such as the death of Michael Jackson, can be seen in short
topic chains. Some short-term issues can be embedded within a long topic chain
because they are related to general topics. One example of such issue is the recall
of Toyota cars which is related to the general topic of the automobile industry.
Those issues embedded within general topics can be identified by looking at focus
shifts shown by words that change significantly within the topic chain.
Our contributions can be summarized as follows:
– We compare six frequently used similarity metrics using log likelihood of
data for finding similar topics. We show that the two most frequently used
metrics, cosine similarity and KL divergence, do not give the best results.
– We define and construct topic chains using the best similarity metric we
found. We then illustrate how to further analyze topic chains to identify
general topics and short-term issues.
– Overall, we propose a framework for understanding how topics and issues
emerge, evolve, and disappear through time in a corpus of online news articles. This framework includes a set of analyses for a sequential corpus that
other similar tools do not provide.
2
Related Work
This work can be positioned with respect to three related research areas: topic
and event detection and tracking, probabilistic topic modeling, and temporal
news mining.
Topic detection and tracking (TDT) is a well-studied task, summarized in
Allan’s book [1], and followed up by a line of research around event threading [5–
7]. Both TDT and event threading solve a narrowly defined problem of looking
for articles related to one or more events, where an event is defined as something
that happens at a certain place at a certain time in the real world. We solve
a much broader problem of discovering all topics and issues that occur in the
corpus, whether or not they are directly related to concrete events in the world.
Also, our definition of issues is more general than the definition of events by the
TDT task. For example, the H1N1 influenza issue of 2009 is a series of related
events such as deaths, vaccinations, and travel warnings, as well as non-events
such as the safety of the vaccine, spatiotemporal course of the pandemic, and
susceptibility of populations. We borrow two central aspects of the TDT task
which are the discovery of new events and the evolution of events over time.
We substitute our general definition of topics for their events such that our
framework discovers new topics and how they evolve over time.
Probabilistic topic models [8] such as the frequently used latent dirichlet
allocation (LDA) [4] discover all topics, regardless of event-like characteristics,
that are highly represented in a corpus, and extensions to LDA, [9, 2, 10, 3] consider the temporal aspect of the corpus as well. In [9], Wang et al. worked with
asynchronous text streams to find common topics from documents with different timestamps. They found highly discriminative topics from asynchronous
data and synchronized the documents according to topics. With dynamic topic
models (DTM) [2], Blei and Lafferty analyzed how topics evolve over time in
a sequential corpus, and they demonstrated how topics in the journal Science
changed from 1881 to 1999. One limitation with DTM is that it only models the
changes of word distributions within the topics and assumes the set of topics
stays constant throughout the corpus, so it does not model how topics appear
and disappear over time. The same limitation exists for the topic trend detection
in [10]. With Topics over Time (TOT) [3], Wang and McCallum jointly model
topics and timestamps to analyze when in the sequential data the topics occur.
This model can discover when new topics appear and then disappear, but in
this model, the topics stay the same over time. In our framework, different but
similar topics form a topic chain so we can observe how the topics evolve over
time.
Previous work on temporal news mining include [11–13]. Leskovec et al. [11]
look at the news cycle by tracking how memes travel widely through the media
sites and blogs. While this approach is very interesting, it does not capture the
broad and overall picture of what topics and issues emerge and spread through
the media sites. Shahat and Guestrin’s work [12] looks at how two news articles
can be connected through a series of articles in between them to form a coherent
chain of articles. This is an effective solution to get a big picture of the story that
connects two news articles. Mei and Zhai’s work [13] is probably the closest to
our work, but they work with data that is filtered for specific topics, such as the
Asia Tsunami. They extend this work in [14] to include the spatial dimension.
Our work aims to present an overall picture of topics and issues including how
to identify general topics as well as temporal issues.
3
Overall Framework
Suppose there is a corpus of twelve months of news articles from major online
newspapers that a user wishes to understand. A good way to do that is to break
down the problem into finding the following details about the corpus:
– Topic: a topic is a major subject discussed in the corpus. Examples are
“winter olympics”, “healthcare reform”, “the stock market”.
– Long-Term Topic: if a topic lasts for a long time, we say it is a long-term
topic. Examples are “the stock market”, “Afghanistan war”, “education”.
– Temporary issue: if a topic lasts for a short time, we say it is a temporary
issue. Examples are “the winter olympics”, “earthquake in Haiti”, “death of
Michael Jackson”.
– Focus Shift: a topic chain exhibits different focuses for each individual topic
in the chain. An example of a focus shift is “Greece, moratorium” to “Europe,
recession” in the “economy” long-term topic.
We propose a framework to analyze the corpus to find the topics, long-term
topics, temporary issues, and focus shifts. In this section, we explain the parts
that compose the overall framework.
1. Discovering Topics: We discover the topics in the corpus with latent dirichlet allocation (LDA) [4], the most widely used method of probabilistic topic
modeling. LDA models topics as multinomial distributions of words.
2. Measuring Topic Similarity: We compare several methods for measuring
topic similarity so that we can use the best method to find similar topics.
We look at six popular similarity metrics and compare them in terms of log
likelihood of data.
3. Constructing Topic Chains: A topic chain is a sequence of similar topics
through time. Using the topic similarity metric, we look for similar topics
within a sliding time window and add links between two similar topics to
construct topic chains.
4. Long-Term Topics and Temporary Issues: After constructing the topic
chains, we can identify long-term topics such as the stock market, temporary
issues such as the Olympics. We can also identify focus shifts in long-term
topics.
4
Topics
The first step in our analysis is finding topics in the corpus. Because we are
looking at news data which are sequential by nature, we divide the corpus into
several time slices, and for each time slice, we find a set of topics that are most
salient in the documents within the time slice. We first describe the topic model
we used for finding the topics, then we describe our dataset and the topics found
in it.
4.1
Latent Dirichlet Allocation
LDA [4] is a widely used method for probability topic modeling. LDA is a generative model that models a document using a mixture of topics. In the generative
process, for each document d, a multinomial distribution θd over topics is randomly sampled from a Dirichlet with parameter α, and then to generate each
word, a topic zn is chosen from this topic distribution, and a word, wn , is generated by randomly sampling from a topic-specific multinomial distribution φzn .
A topic-specific multinomial distribution φzn is also randomly sampled from a
Table 1. Four topics discovered by LDA for the news dataset. Topics are randomly
chosen and are represented by top ten probability words. Topic 1 is about “soccer
game”, topic 2 is about “market” and “business”, topic 3 is about “smart phones”,
and topic 4 is about “research”. Each topic is a multinomial distribution over words.
Topic 1
Top words Probability
game
0.030
player
0.026
league
0.025
coach
0.023
soccer
0.016
season
0.012
leader
0.011
competition
0.011
advance
0.007
pro
0.007
Topic 2
Topic 3
Top words Probability Top words Probability
growth
0.035
Apple
0.024
business
0.034
smartphone
0.018
recovery
0.031
internet
0.017
crisis
0.026
iphone
0.016
prospect
0.024
mobile phone
0.013
policy
0.023
Google
0.012
investment
0.020
computer
0.011
strategy
0.018
usage
0.010
market
0.016
advertise
0.010
consume
0.015
information
0.008
Topic 4
Top words Probability
research
0.078
professor
0.042
science
0.018
doctorate
0.017
discovery
0.016
analysis
0.012
technology
0.010
universe
0.010
plant
0.009
experiment
0.009
Dirichlet with parameter β. From the generative process, we obtain the likelihood
of a document:
p(w, z, θd , Φ|α, β)
=
N
Y
p(wn |φzn )p(zn |θd ) · p(θd |α) · p(Φ|β).
n=1
The Dirichlet parameters α and β are vectors that represent the average of the
respective distributions. In many applications, it is sufficient to assume that such
vectors are uniform and to fix them at a value pre-defined by the user, and these
values act as smoothing coefficients.
4.2
Corpus
We collected over 130K news documents from the Web editions of three major
Korean newspapers1 between 2009-07-01 and 2010-04-10. Each news outlet covers a wide range of topics such as politics, economy, sports, entertainment, and
culture, and show their own perspectives on cultural and social phenomena.
For the topic modeling task, we refined each document using a Korean morpheme analyzer and part-of-speech (POS) tagger provided by ETRI2 . In the
Korean language, each word can be broken down into morphemes. The morphemes are the smallest meaningful units, and each morpheme has a POS tag
associated with it. Most of the morphemes do not carry semantic meaning but
are instead used as syntactic markers, and almost every verb, adverb, and adjective can be broken down into morphemes with a noun token and one or more
syntactic markers.
After preprocessing the documents as described, we divided the corpus into
28 time slices, ten days each. The average number of documents in each time
1
2
http://www.yonhapnews.co.kr/, http://www.donga.com/, http://www.hani.co.kr/
http://www.etri.re.kr
slice is 4,715, and the average number of unique words in each group is 13,611.
We extracted 50 topics with LDA for every time slice using Gibbs sampling. To
reduce the effort of estimating hyperparameters, we used symmetric Dirichlet
priors. More specifically, for α and β, we adopted the commonly used values of
0.1 and 0.01 respectively. We set the number of topics to be 50 for one time slice,
so the total number of topics is 1,400 for the entire corpus. We randomly chose
4 topics from the corpus and show them in Table 1, each topic represented with
the words that have the highest probabilities in that topic.
5
Topic Similarity
To construct topic chains, we need to measure the similarity between a pair
of topics. In previous topic modeling research where topic similarity must be
measured, cosine similarity [15] and Kullback-Leibler (KL) divergence [16] are
frequently used without any formal validation. There exist, however, several wellknown metrics that can be used to measure topic similarity, so we compared them
to see which metric would be best for our purpose. We considered six metrics
and evaluated each metric using the negative log likelihood of corpus.
5.1
Six Metrics of Topic Similarity
A topic, φi , is a multinomial distribution over the vocabulary, but it can also
be viewed as a ranked list of words, or a |W | dimensional vector, where each
dimension i is a probability of wi in that topic. A topic can also be represented
by a set of topic words–words with a probability over a threshold. These various
perspectives allow the following metrics for measuring similarity between topic
φi and topic φj :
– Cosine Similarity measures the similarity between two vectors by finding
the cosine of the angle between them.
– Jaccard’s Coefficient measures the similarity and diversity of two sets. It
is defined as the size of the intersection divided by the size of the union of
two sets.
– Kendall’s τ Coefficient measures the association between two ranked lists.
– Discounted Cumulative Gain(DCG) measures the effectiveness of the
ranked results of a web search algorithm.
– Kullback-Leibler Divergence is a non-symmetric measure of the difference between two probability distributions p and q.
– Jensen-Shannon Divergence is the symmetric variation of KL divergence.
Each metric considers a different aspect of the relationship between two topics. Kendall’s τ and DCG consider the ranks of words within a topic. KL divergence and JS divergence consider the divergence of multinomial topic probabilities, and lower divergence would indicate higher similarity between two topics.
Cosine similarity measures the angle of two vectors, and Jaccard’s coefficient
1945000
Negative Log Likelihood
1935000
1925000
1915000
1905000
1895000
1885000
D
C
G
*
JS
*
KL
e
in
os
C
0.
5
cc
Ja
ar
cc
ar
d-
d-
0.
3*
.5
*
-0
al
Ja
nd
Ke
Ke
nd
al
-0
.3
*
1875000
Similarity Metrics
Fig. 1. Comparison of negative log likelihood for six similarity metrics using a boxplot.
Negative log likelihood was computed for the corpus using the set of topics where
five topics were substituted with the five most similar topics from another time slice,
identified by each of the six similarity metrics. A better similarity metric gives a lower
negative log likelihood. JS divergence and Jaccard’s coefficient with 0.5 cumulative
probability mass achieve better performances than the other metrics. An asterisk (*)
next to a metric indicates statistically significant differences between the metric and
JS divergence using t-test, p < 0.01.
looks at the association between two sets. Jaccard’s coefficient must use a partial set of words because it looks at the intersection and the union of the two
sets of words that represent the topics. We use the top probability words that
contribute to the cumulative probability mass, which is a parameter that must
be set. We also use a partial set of top probability words for Kendall’s τ . This is
because Kendall’s τ is equally affected by the differences among high probability
words and the differences among low probability words, but the words that have
low probabilities in both topics should not contribute to the similarity score as
much.
5.2
Comparing the Metrics
We compared the six metrics with the negative log likelihood of the corpus which
measures how well the model explains the corpus. Starting from a set of topics
extracted for a time slice, we substitute five topics with the topics from another
time slice that are found to be most similar according to each of the six metrics
to form six modified sets of topics. By comparing the negative log likelihoods
using the modified sets of topics, we can see which metrics found the most similar
topics. The process is as follows:
1. Train LDA for two consecutive time slices to get two sets of topics Φt−1 =
t−1
t−1
t−1
t
t
t
t
t
{φt−1
1 , φ2 , φ3 , ...φk } and Φ = {φ1 , φ2 , φ3 , ...φk }.
2. Compute the similarity score between φit−1 and φtj for every i, j.
3. Select top five pairs of similar topics from the two topic sets.
4. Substitute the original topics Φt = {φt1 , φt2 , φt3 , ...φtk } with the five most sim(t−1)
ilar topics from t − 1. So the Φtnew = {φt1 , φi
, φt3 , ...φtk }, where i is a one
of the five most similar topics from the previous time slice.
5. Finally, using Φtnew , calculate the log-likelihood of data at time t.
To evaluate the metrics, we selected the first two consecutive time slices,
and then trained LDA on each time slice 30 times. Using these 30 pairs of
LDA results, we calculated the similarities of all topic pairs, replaced the most
similar topics, and computed the negative log likelihoods. As Figure 1 shows, JS
divergence and Jaccard’s Coefficient produced the lowest log likelihood scores,
which we interpret to mean they performed the best among the six metrics.
As we noted before, Jaccard’s coefficient and Kendall’s τ use a subset of the
vocabulary–top probability words that contribute to a cumulative probability
mass. The average size of the set of words with probability mass 0.5 is 39.56,
and 0.3 is 13.58. The results show that Jaccard’s coefficient can find similar
topics at probability mass of 0.5, using only the top 40 words. Kendall’s τ does
not show good performance compared to Jaccard’s coefficient although they use
the same set of words. This result indicates that the ranking of top probability
words does not matter much in judging topic similarity. DCG does not perform
well for this topic similarity task even though it is a good metric of comparing
ranked results in information retrieval (IR). This is because the typical results of
IR include relevance scores, but the topics found by LDA do not have analogous
scores to be used in place of relevance scores.
We further tested Jaccard’s Coefficient with various probability masses. However, selecting a proper probability mass can be corpus-dependent. Hence, we
conclude that JS divergence is best in terms of performance and generality, so
we use JS divergence as the topic similarity metric in the rest of the paper.
6
Topics and Issues
Using the similarity discussed in the previous section, we construct topic chains
to understand the topic trends in the main stream news. In this section, we
discuss the construction of topic chains and associated parameters, interpretation
of long topic chains, and the characteristics of short topic chains.
6.1
Constructing Topic Chains
We construct topic chains by finding similar topics within a certain time window. We use two parameters, similarity cut and sliding window, and follow this
process:
1. Calculate the similarity between topic φti and topic φjt−1 for all topics at
time t − 1.
200
Window Size 1
Window Size 3
Window Size 6
Number of Topic Chains
175
150
125
100
75
50
25
0
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
JS Divergence Cut
Fig. 2. Number of topic chains with different similarity cuts using JS divergence. The
number of topic chains is significantly changed at JS divergence of 0.4. We chose JS
divergence of 0.4 to construct topic chains.
2. If there are one or more topics such that sim(φti , φt−1
j ) is greater than the
similarity cut, we make links between all such topic pairs, and move to the
next topic φti+1 .
3. If there are no similar topic pairs, we calculate similarity between φti and
Φt−2
4. Repeat, going back one more time slice, until one or more similar topics
are found, or the time gap between the two time slices exceeds the sliding
window size.
The two parameters, similarity cut and the window size, play important roles
in determining the characteristics of the topic chains constructed. We discuss
each of them below.
Similarity Cut There is no standard similarity cut at which we can say two
topics are similar, so we construct several topic chains, varying the similarity
cut and looking at the effect on the resulting topic chains. Figure 2 shows how
the number of topic chains changes with similarity cut using JS divergence. We
define the size of a topic chain to be the number of topics in that chain, and we
count topic chains whose size is greater than one. We also experimented with
various sizes of the sliding window. If we set the JS divergence cut to a large
value, then all topic nodes would be disconnected, and the total number of topic
chains of size greater than one would be 0. Conversely, if we set the JS divergence
cut to 0, then all topic nodes would be connected, and the number of topic chains
would be 1. As Figure 2 shows, the number of topic chains changes significantly
at 0.4. To see the relationship between JS divergence values and the similarity
of two topics in a qualitative way, we can look at pairs of topics and the JS
divergence values. From the qualitative analysis and the analysis of the number
of topic chains, we decided that 0.4 is an appropriate threshold of JS divergence
for constructing topic chains.
07M 0W 34T
78.26
07M 0W 21T
86.41
07M 0W 33T
70.03
07M 1W 1T
130.4
07M 1W 4T
113.6
07M 1W 5T
102.9
07M 2W 7T
229.5
07M 2W 10T
184.2
07M 2W 30T
184.9
07M 0W 38T
75.02
08M 0W 14T
102.0
07M 0W 8T
108.5
07M 1W 9T
170.1
07M 1W 10T
174.5
07M 2W 4T
307.2
07M 2W 49T
291.1
08M 0W 1T
120.7
08M 1W 41T
138.8
07M 0W 27T
117.0
08M 2W 8T
261.1
09M 0W 35T
234.5
09M 1W 48T
172.4
09M 1W 8T
237.7
09M 2W 17T
187.6
10M 1W 35T
134.4
07M 0W 10T
112.0
07M 1W 46T
135.1
07M 2W 45T
235.2
08M 0W 21T
151.6
08M 0W 37T
114.5
10M 0W 20T
494.0
10M 2W 32T
353.4
11M 0W 29T
223.7
11M 1W 19T
279.5
07M 0W 48T
82.50
07M 1W 23T
124.6
07M 2W 9T
174.3
12M 0W 13T
227.9
12M 0W 26T
296.0
12M 1W 3T
214.1
12M 1W 42T
320.8
10M 2W 17T
299.6
12M 0W 8T
216.0
12M 1W 17T
204.9
11M 1W 45T
164.0
11M 2W 46T
167.5
11M 2W 21T
151.6
12M 0W 17T
162.1
12M 1W 29T
202.6
12M 2W 45T
177.6
12M 1W 1T
174.6
12M 2W 41T
255.8
12M 1W 48T
129.9
12M 2W 12T
166.6
07M 1W 34T
170.9
07M 0W 17T
60.64
08M 0W 29T
81.55
07M 0W 31T
93.05
07M 0W 6T
99.31
07M 1W 40T
116.0
07M 1W 41T
118.4
07M 2W 32T
161.0
07M 2W 13T
211.3
08M 0W 44T
111.0
08M 0W 17T
151.3
08M 0W 13T
137.7
08M 1W 48T
109.1
08M 1W 10T
150.4
08M 1W 40T
160.1
08M 1W 45T
110.5
08M 2W 14T
229.3
08M 2W 40T
141.4
08M 2W 30T
188.8
07M 1W 44T
145.6
07M 2W 15T
246.4
08M 2W 42T
209.3
09M 0W 0T
153.4
09M 0W 13T
197.9
09M 1W 42T
187.0
09M 2W 24T
221.6
07M 0W 21T
86.41
07M 0W 33T
70.03
07M 1W 4T
113.6
07M 1W 5T
102.9
07M 2W 10T
184.2
07M 2W 30T
184.9
12M 0W 6T
149.7
12M 0W 4T
141.9
12M 0W 10T
187.3
12M 1W 12T
181.6
12M 1W 16T
130.5
12M 1W 40T
209.7
12M 0W 28T
193.4
12M 1W 24T
193.7
12M 2W 8T
143.3
12M 2W 4T
216.4
01M 0W 13T
41.88
07M 0W 8T
108.5
07M 0W 27T
117.0
07M 1W 10T
174.5
07M 2W 3T
332.0
07M 0W 33T
70.03
09M 0W 7T
289.7
09M 1W 37T
337.3
10M 1W 3T
192.1
10M 2W 1T
348.6
10M 2W 13T
225.7
11M 0W 19T
257.8
11M 0W 9T
176.9
11M 0W 23T
211.4
11M 1W 19T
279.5
11M 1W 42T
362.0
11M 1W 41T
214.2
11M 1W 20T
220.0
12M 0W 1T
287.5
11M 2W 27T
205.5
12M 1W 6T
210.9
12M 1W 17T
204.9
11M 2W 39T
227.3
11M 2W 47T
341.4
12M 0W 13T
227.9
07M 1W 9T
170.1
07M 2W 4T
307.2
12M 0W 11T
143.8
08M 0W 32T
134.9
08M 0W 7T
121.9
08M 1W 25T
197.8
08M 1W 26T
101.4
08M 2W 41T
192.4
08M 0W 24T
185.5
08M 1W 27T
233.2
08M 2W 6T
312.3
09M 0W 30T
220.9
10M 1W 36T
206.6
10M 2W 22T
274.3
11M 0W 21T
158.6
08M 0W 26T
186.5
08M 1W 14T
255.7
08M 2W 25T
295.3
09M 0W 20T
261.8
09M 1W 27T
250.0
09M 2W 18T
356.0
10M 0W 24T
521.2
08M 0W 46T
113.7
08M 1W 47T
134.9
08M 1W 3T
105.6
08M 1W 31T
126.6
08M 1W 34T
141.3
08M 1W 4T
129.2
08M 1W 36T
239.4
08M 1W 38T
158.8
08M 2W 0T
170.3
08M 2W 2T
150.4
08M 2W 3T
154.6
08M 2W 11T
186.5
08M 2W 27T
177.4
08M 2W 33T
277.5
08M 2W 15T
204.5
09M 0W 22T
214.9
09M 0W 29T
224.4
08M 1W 1T
200.8
08M 2W 12T
152.2
08M 2W 46T
122.3
08M 2W 34T
198.2
08M 2W 45T
231.2
09M 0W 6T
159.3
09M 0W 26T
133.4
09M 0W 44T
184.3
09M 0W 41T
237.4
09M 1W 21T
138.6
09M 1W 49T
210.5
09M 1W 25T
199.3
09M 2W 6T
248.2
09M 2W 1T
175.0
09M 2W 14T
269.7
10M 0W 7T
359.8
10M 0W 10T
289.3
10M 0W 47T
287.4
09M 0W 36T
197.3
09M 0W 37T
164.8
09M 0W 18T
137.9
09M 0W 23T
146.5
09M 0W 31T
164.4
09M 0W 3T
195.5
09M 0W 15T
184.4
09M 1W 3T
255.7
09M 1W 18T
176.3
09M 1W 22T
143.6
09M 1W 24T
165.9
09M 1W 28T
180.1
09M 1W 34T
177.7
09M 1W 40T
189.2
09M 2W 9T
214.8
09M 2W 0T
200.3
09M 2W 26T
244.3
09M 2W 16T
154.6
09M 1W 26T
149.5
10M 0W 37T
379.5
10M 0W 36T
209.3
10M 1W 12T
106.2
11M 0W 28T
119.4
09M 2W 23T
254.6
10M 0W 32T
492.1
10M 1W 29T
135.9
10M 1W 16T
233.7
11M 0W 42T
140.3
11M 0W 14T
133.5
11M 1W 15T
140.7
09M 1W 4T
231.2
09M 1W 43T
156.4
09M 1W 44T
292.8
09M 1W 9T
241.1
09M 2W 35T
174.8
09M 2W 42T
155.5
09M 2W 45T
399.7
09M 2W 49T
276.3
12M 2W 25T
160.8
01M 0W 34T
61.04
11M 2W 12T
305.4
02M 1W 14T
219.1
01M 1W 2T
186.6
02M 2W 14T
177.1
01M 2W 48T
165.2
02M 0W 38T
145.3
09M 2W 32T
188.7
09M 2W 41T
252.2
09M 2W 44T
248.0
10M 0W 22T
370.1
10M 0W 26T
212.7
10M 0W 27T
393.7
10M 0W 28T
333.6
10M 1W 47T
254.5
10M 1W 41T
121.2
10M 0W 33T
239.3
10M 1W 28T
154.3
10M 0W 35T
305.0
10M 0W 2T
321.5
10M 0W 17T
379.9
10M 0W 30T
350.6
10M 1W 22T
196.9
10M 1W 38T
168.3
10M 1W 40T
186.6
10M 1W 44T
219.8
10M 2W 12T
213.4
10M 2W 0T
305.9
10M 2W 15T
183.2
10M 2W 34T
377.8
10M 2W 36T
238.1
11M 0W 2T
119.1
11M 0W 40T
205.7
10M 1W 2T
189.5
10M 1W 37T
262.5
10M 1W 20T
232.2
10M 1W 4T
162.1
10M 1W 13T
170.8
10M 2W 4T
183.0
10M 2W 10T
250.3
10M 2W 44T
227.4
10M 2W 41T
241.8
10M 2W 28T
220.4
11M 0W 35T
134.7
10M 2W 8T
220.7
10M 2W 20T
291.7
10M 2W 23T
274.6
10M 2W 35T
185.3
10M 2W 45T
162.8
11M 0W 4T
132.5
11M 0W 22T
293.6
11M 0W 26T
124.0
11M 0W 36T
217.2
11M 0W 49T
159.1
11M 1W 3T
194.6
11M 0W 47T
145.7
11M 0W 45T
123.1
11M 1W 1T
140.9
11M 1W 11T
185.0
11M 1W 35T
179.9
11M 0W 0T
176.8
11M 0W 48T
240.6
11M 1W 7T
208.1
11M 1W 28T
324.4
11M 2W 37T
179.4
11M 2W 11T
257.4
11M 2W 40T
159.4
12M 0W 3T
260.2
11M 0W 13T
164.2
11M 1W 22T
190.5
11M 1W 27T
201.7
11M 1W 40T
213.1
11M 2W 6T
164.8
11M 2W 17T
163.4
11M 1W 9T
291.3
11M 2W 24T
287.0
12M 0W 47T
211.0
11M 1W 18T
140.1
11M 1W 46T
169.0
12M 0W 31T
178.2
12M 1W 36T
200.0
11M 2W 13T
131.0
11M 2W 28T
142.7
12M 1W 43T
120.0
12M 2W 0T
202.6
11M 2W 5T
159.2
11M 1W 10T
268.9
11M 1W 12T
198.1
11M 1W 49T
273.2
11M 1W 13T
275.7
11M 1W 21T
231.4
11M 2W 30T
211.0
11M 2W 32T
236.6
11M 2W 34T
228.5
11M 2W 35T
228.7
11M 2W 43T
218.1
11M 2W 49T
204.5
12M 0W 29T
203.5
11M 2W 23T
173.9
12M 0W 7T
215.2
12M 0W 20T
226.8
12M 1W 49T
210.6
12M 1W 15T
162.7
12M 1W 44T
193.8
12M 2W 49T
152.4
12M 2W 17T
217.5
12M 2W 37T
231.9
11M 2W 10T
155.8
11M 2W 33T
145.1
11M 2W 19T
271.9
12M 0W 11T
143.8
12M 0W 25T
155.6
12M 0W 42T
233.2
12M 1W 18T
140.6
12M 0W 18T
143.1
12M 0W 27T
112.5
12M 0W 41T
191.6
12M 0W 15T
216.0
12M 0W 46T
159.3
12M 1W 10T
145.5
12M 1W 14T
142.5
12M 1W 20T
194.5
12M 1W 25T
231.2
12M 1W 27T
156.8
12M 2W 34T
154.0
12M 0W 40T
280.6
12M 0W 23T
179.7
12M 0W 9T
196.5
12M 1W 30T
264.8
12M 1W 31T
123.9
12M 1W 32T
177.3
12M 1W 33T
201.5
12M 2W 7T
170.9
12M 2W 16T
200.5
12M 2W 20T
307.0
12M 2W 13T
160.4
12M 0W 21T
204.2
01M 0W 9T
112.1
01M 0W 45T
78.13
12M 0W 30T
121.9
12M 0W 48T
308.4
12M 0W 24T
140.6
12M 1W 2T
143.0
12M 1W 7T
178.1
12M 1W 47T
112.1
12M 1W 0T
160.0
12M 1W 34T
149.4
12M 1W 41T
260.4
12M 1W 46T
169.1
12M 2W 21T
160.8
12M 2W 27T
155.3
12M 2W 39T
112.0
12M 2W 46T
122.8
12M 2W 48T
235.5
12M 2W 11T
198.3
12M 2W 33T
133.9
12M 2W 2T
126.6
01M 0W 38T
40.48
01M 0W 11T
102.4
01M 0W 16T
62.12
01M 0W 26T
72.61
01M 1W 3T
49.54
12M 2W 5T
132.3
12M 2W 30T
161.5
01M 0W 32T
65.18
01M 0W 42T
61.69
01M 2W 40T
204.8
01M 1W 7T
81.27
01M 1W 42T
81.03
02M 0W 33T
195.3
01M 0W 18T
56.21
01M 1W 5T
78.32
01M 1W 32T
154.5
01M 0W 37T
41.30
01M 0W 19T
108.6
01M 1W 14T
74.30
01M 1W 23T
178.5
01M 0W 23T
54.77
01M 1W 33T
101.3
01M 2W 47T
76.50
02M 1W 15T
145.5
01M 0W 39T
102.9
01M 1W 41T
115.6
01M 2W 32T
102.1
02M 0W 26T
74.02
01M 0W 49T
148.1
01M 0W 47T
58.68
01M 1W 45T
193.2
01M 1W 49T
89.76
01M 2W 5T
229.9
02M 0W 47T
96.71
02M 1W 23T
113.2
01M 1W 38T
148.0
01M 1W 18T
191.5
01M 2W 12T
191.8
01M 2W 14T
155.8
02M 0W 22T
160.6
02M 0W 37T
143.3
03M 0W 16T
149.4
03M 1W 43T
138.3
03M 0W 17T
151.0
03M 1W 37T
163.7
01M 1W 31T
61.66
01M 2W 36T
140.8
01M 1W 29T
141.6
01M 2W 26T
81.10
02M 0W 43T
73.85
02M 1W 33T
219.0
02M 1W 31T
162.9
02M 2W 23T
217.4
02M 2W 17T
149.3
03M 0W 15T
127.6
01M 1W 12T
160.7
01M 2W 10T
171.3
02M 0W 16T
146.1
02M 0W 18T
193.3
02M 1W 42T
170.4
02M 2W 4T
132.7
03M 0W 6T
150.7
02M 2W 10T
186.0
02M 2W 16T
197.0
03M 0W 7T
194.3
03M 1W 28T
190.5
01M 1W 46T
108.9
01M 1W 36T
113.2
01M 2W 30T
116.3
01M 2W 31T
120.7
01M 2W 41T
92.82
02M 0W 23T
138.4
02M 0W 15T
80.65
02M 1W 25T
204.5
02M 1W 22T
132.0
02M 2W 37T
159.7
02M 2W 18T
167.7
02M 2W 31T
126.1
03M 1W 41T
158.9
03M 1W 35T
211.2
03M 0W 14T
139.6
03M 0W 18T
164.4
01M 1W 11T
104.5
01M 1W 17T
106.1
01M 2W 44T
139.2
01M 2W 4T
224.8
03M 2W 38T
194.8
03M 2W 4T
179.7
03M 2W 11T
145.0
03M 2W 22T
176.0
03M 2W 45T
158.1
03M 2W 1T
170.3
04M 0W 27T
58.27
04M 0W 30T
83.01
04M 0W 15T
57.49
04M 0W 42T
66.10
04M 0W 16T
67.07
04M 0W 26T
63.00
02M 0W 30T
105.1
02M 0W 27T
55.77
02M 0W 46T
72.83
02M 1W 2T
114.1
02M 1W 5T
149.2
02M 1W 6T
139.9
02M 1W 8T
86.81
02M 1W 27T
165.7
02M 2W 38T
112.1
02M 0W 6T
66.62
02M 2W 30T
128.9
02M 2W 35T
120.3
02M 0W 19T
169.7
02M 0W 1T
110.2
02M 0W 48T
141.5
03M 0W 21T
163.8
03M 1W 17T
178.0
03M 0W 22T
194.9
03M 1W 44T
170.7
12M 1W 26T
142.9
12M 2W 15T
165.7
03M 2W 41T
214.7
03M 0W 19T
135.5
02M 1W 1T
160.9
02M 1W 26T
245.6
02M 1W 17T
144.7
02M 2W 2T
168.7
02M 2W 6T
259.8
02M 2W 7T
115.8
03M 0W 5T
176.5
03M 0W 9T
266.7
03M 0W 38T
132.6
03M 1W 19T
277.4
03M 1W 38T
268.1
03M 1W 36T
160.1
03M 1W 46T
159.9
03M 2W 23T
318.1
03M 2W 48T
148.6
04M 0W 28T
122.4
02M 1W 43T
127.3
02M 2W 9T
128.0
02M 1W 39T
160.9
02M 1W 7T
121.4
02M 1W 49T
146.8
02M 2W 15T
123.3
02M 2W 40T
171.6
02M 2W 28T
141.3
02M 2W 29T
130.1
03M 0W 26T
140.1
02M 1W 44T
163.6
03M 0W 30T
218.3
03M 1W 47T
199.9
03M 2W 35T
173.9
04M 0W 18T
120.3
03M 2W 27T
198.7
03M 2W 32T
191.6
04M 0W 23T
85.85
04M 0W 29T
84.74
02M 1W 32T
143.6
02M 1W 11T
155.2
02M 1W 46T
189.8
02M 1W 18T
201.6
02M 1W 28T
180.7
02M 1W 38T
167.6
02M 1W 35T
226.6
02M 2W 32T
156.0
02M 2W 34T
177.8
02M 2W 36T
154.0
02M 2W 39T
201.9
02M 2W 49T
214.4
02M 2W 43T
176.8
02M 2W 47T
132.0
02M 2W 48T
185.8
03M 0W 12T
155.4
03M 0W 2T
162.9
03M 0W 1T
201.3
03M 0W 37T
253.4
03M 0W 20T
206.5
03M 1W 39T
117.4
03M 1W 8T
254.4
02M 1W 0T
226.4
02M 2W 41T
113.7
02M 2W 25T
209.4
02M 2W 5T
138.1
03M 0W 23T
136.3
03M 0W 27T
246.7
03M 0W 34T
140.4
03M 1W 45T
163.1
03M 1W 3T
261.0
03M 1W 29T
239.6
03M 1W 10T
249.6
03M 2W 29T
143.0
03M 2W 31T
301.3
03M 2W 42T
241.8
03M 2W 33T
238.3
04M 0W 32T
63.06
02M 2W 33T
112.1
02M 2W 1T
117.0
03M 0W 31T
188.2
03M 0W 48T
184.6
03M 0W 46T
130.6
03M 0W 49T
93.48
03M 1W 11T
175.0
03M 1W 31T
228.3
03M 1W 40T
170.4
03M 1W 5T
128.0
03M 1W 22T
173.9
03M 2W 25T
188.0
03M 0W 40T
123.6
03M 1W 25T
124.2
03M 1W 7T
136.9
03M 1W 33T
201.0
03M 2W 6T
137.2
03M 2W 7T
145.0
03M 2W 19T
124.4
03M 1W 14T
194.6
03M 1W 1T
282.2
03M 2W 9T
183.9
03M 2W 24T
301.4
04M 0W 49T
83.59
04M 0W 7T
104.7
03M 0W 4T
220.8
03M 1W 6T
263.1
02M 0W 17T
83.46
02M 1W 20T
169.0
03M 2W 30T
231.6
02M 1W 48T
170.6
02M 2W 20T
129.4
04M 0W 46T
98.31
03M 0W 1T
201.3
03M 2W 37T
222.7
03M 2W 42T
241.8
02M 2W 49T
214.4
03M 1W 13T
156.6
08M 2W 14T
229.3
07M 2W 1T
198.0
08M 0W 4T
122.9
08M 1W 24T
128.4
08M 2W 48T
171.3
08M 2W 18T
153.0
09M 0W 4T
143.6
09M 1W 15T
136.2
11M 2W 25T
156.9
12M 0W 17T
162.1
11M 2W 9T
159.5
12M 0W 6T
149.7
12M 1W 1T
174.6
12M 1W 48T
129.9
12M 2W 12T
166.6
01M 0W 3T
51.24
01M 0W 28T
60.40
01M 0W 25T
51.30
11M 2W 46T
167.5
12M 0W 37T
138.9
12M 1W 12T
181.6
12M 2W 3T
144.4
11M 2W 21T
151.6
12M 1W 16T
130.5
08M 1W 48T
109.1
08M 2W 30T
188.8
11M 0W 34T
173.4
11M 2W 14T
235.5
12M 0W 2T
133.5
12M 0W 10T
187.3
12M 1W 21T
156.6
12M 2W 9T
121.6
11M 2W 7T
215.5
12M 1W 11T
151.7
07M 0W 8T
108.5
07M 0W 27T
117.0
07M 1W 10T
174.5
07M 1W 24T
226.3
07M 0W 4T
149.1
07M 2W 3T
332.0
07M 2W 25T
312.0
08M 0W 49T
210.2
07M 0W 10T
112.0
07M 0W 13T
95.21
07M 1W 36T
188.0
07M 1W 46T
135.1
09M 0W 7T
289.7
07M 2W 9T
174.3
07M 2W 0T
175.6
08M 0W 33T
114.3
08M 0W 40T
139.9
09M 1W 16T
241.8
10M 1W 3T
192.1
10M 2W 1T
348.6
10M 2W 13T
225.7
09M 2W 21T
224.5
11M 0W 46T
205.9
11M 0W 29T
223.7
11M 0W 19T
257.8
11M 1W 14T
257.4
11M 1W 42T
362.0
11M 1W 41T
214.2
11M 2W 47T
341.4
12M 0W 1T
287.5
11M 2W 27T
205.5
12M 0W 26T
296.0
12M 1W 3T
214.1
12M 1W 42T
320.8
12M 1W 6T
210.9
12M 1W 17T
204.9
09M 1W 13T
162.6
11M 1W 20T
220.0
11M 2W 15T
233.1
09M 2W 29T
201.0
12M 1W 29T
202.6
12M 2W 41T
255.8
01M 1W 38T
148.0
01M 2W 10T
171.3
09M 0W 2T
166.2
09M 1W 7T
186.5
09M 2W 36T
209.8
10M 0W 18T
243.6
10M 0W 42T
301.0
10M 1W 43T
163.0
10M 0W 9T
273.2
10M 2W 33T
225.6
10M 2W 39T
249.7
10M 1W 24T
167.2
07M 0W 29T
52.55
02M 2W 17T
149.3
03M 0W 18T
164.4
07M 1W 2T
180.1
07M 2W 20T
249.4
07M 2W 21T
234.8
08M 0W 3T
139.2
08M 1W 16T
166.0
07M 1W 22T
143.5
07M 1W 31T
225.6
07M 2W 22T
211.6
07M 2W 24T
338.7
08M 0W 8T
135.0
08M 1W 5T
139.8
07M 1W 14T
128.9
07M 1W 7T
147.2
07M 1W 30T
198.2
07M 2W 35T
178.8
07M 2W 40T
292.3
08M 1W 39T
136.7
08M 2W 44T
230.2
09M 0W 34T
219.3
08M 2W 43T
192.3
09M 1W 41T
137.8
09M 2W 44T
248.0
10M 0W 28T
333.6
07M 2W 31T
174.0
08M 0W 0T
120.9
08M 1W 2T
148.1
09M 0W 18T
137.9
09M 1W 22T
143.6
10M 1W 13T
170.8
10M 2W 45T
162.8
07M 1W 48T
173.7
07M 2W 38T
190.3
07M 1W 47T
162.4
07M 2W 48T
211.8
07M 2W 44T
267.8
08M 0W 9T
118.2
08M 0W 11T
157.8
08M 0W 25T
193.1
08M 0W 12T
114.5
08M 1W 0T
152.1
08M 0W 32T
134.9
07M 2W 23T
157.2
08M 0W 20T
153.4
08M 1W 3T
105.6
08M 1W 13T
92.15
08M 1W 25T
197.8
08M 2W 22T
247.0
08M 2W 2T
150.4
09M 0W 45T
155.5
08M 2W 41T
192.4
09M 0W 10T
201.9
09M 0W 30T
220.9
09M 1W 4T
231.2
09M 1W 10T
235.2
09M 2W 35T
174.8
09M 2W 47T
274.3
10M 1W 27T
132.3
08M 2W 7T
160.7
09M 1W 2T
215.2
10M 0W 34T
362.1
09M 1W 45T
187.1
09M 2W 12T
208.1
10M 1W 36T
206.6
10M 0W 6T
314.4
10M 2W 22T
274.3
10M 1W 6T
152.7
10M 1W 32T
152.9
10M 2W 16T
240.5
11M 0W 15T
150.5
07M 2W 46T
163.9
08M 1W 29T
180.8
08M 2W 28T
256.3
08M 1W 11T
155.4
09M 0W 17T
161.4
09M 1W 14T
179.8
09M 2W 20T
189.8
10M 0W 48T
258.6
10M 2W 42T
178.7
08M 0W 7T
121.9
12M 0W 22T
156.1
12M 0W 12T
188.7
12M 1W 9T
140.8
12M 2W 19T
175.6
09M 2W 14T
269.7
10M 0W 47T
287.4
08M 0W 46T
113.7
08M 1W 47T
134.9
08M 1W 36T
239.4
08M 1W 38T
158.8
08M 2W 0T
170.3
08M 2W 33T
277.5
08M 2W 15T
204.5
08M 1W 31T
126.6
08M 2W 3T
154.6
08M 1W 34T
141.3
09M 0W 23T
146.5
08M 1W 1T
200.8
08M 2W 11T
186.5
08M 2W 46T
122.3
09M 1W 43T
156.4
09M 0W 36T
197.3
09M 1W 3T
255.7
09M 2W 26T
244.3
10M 0W 37T
379.5
10M 0W 36T
209.3
10M 1W 12T
106.2
11M 0W 28T
119.4
11M 2W 43T
218.1
08M 2W 5T
255.6
09M 0W 37T
164.8
09M 0W 31T
164.4
09M 0W 3T
195.5
09M 1W 9T
241.1
09M 1W 18T
176.3
09M 1W 28T
180.1
09M 1W 34T
177.7
09M 2W 49T
276.3
09M 2W 9T
214.8
10M 1W 34T
239.0
10M 2W 41T
241.8
10M 1W 31T
155.1
11M 0W 36T
217.2
10M 2W 24T
167.4
11M 1W 13T
275.7
12M 0W 21T
204.2
09M 0W 44T
184.3
09M 1W 21T
138.6
09M 2W 16T
154.6
10M 0W 7T
359.8
10M 2W 30T
305.7
11M 1W 49T
273.2
08M 2W 34T
198.2
09M 0W 26T
133.4
09M 2W 42T
155.5
09M 2W 6T
248.2
10M 1W 20T
232.2
12M 1W 33T
201.5
09M 0W 6T
159.3
09M 0W 41T
237.4
09M 1W 49T
210.5
09M 2W 40T
281.5
10M 2W 23T
274.6
11M 2W 35T
228.7
08M 2W 12T
152.2
08M 2W 45T
231.2
09M 1W 24T
165.9
09M 0W 29T
224.4
09M 1W 25T
199.3
09M 2W 1T
175.0
10M 0W 10T
289.3
11M 1W 37T
330.5
11M 0W 14T
133.5
09M 0W 15T
184.4
10M 1W 2T
189.5
10M 2W 8T
220.7
09M 1W 44T
292.8
09M 2W 23T
254.6
09M 2W 11T
203.8
09M 2W 45T
399.7
11M 2W 37T
179.4
10M 1W 7T
227.8
09M 2W 3T
174.8
11M 2W 1T
111.2
12M 0W 38T
139.6
09M 2W 32T
188.7
09M 2W 41T
252.2
10M 0W 21T
346.7
10M 0W 26T
212.7
10M 0W 27T
393.7
10M 0W 33T
239.3
10M 1W 41T
121.2
10M 0W 30T
350.6
10M 1W 44T
219.8
10M 2W 36T
238.1
11M 0W 2T
119.1
11M 1W 29T
230.9
11M 0W 8T
137.5
09M 2W 43T
216.4
10M 1W 37T
262.5
10M 0W 14T
256.6
10M 2W 4T
183.0
10M 1W 45T
155.8
10M 2W 20T
291.7
10M 2W 44T
227.4
11M 0W 4T
132.5
11M 1W 9T
291.3
11M 0W 26T
124.0
11M 1W 1T
140.9
11M 2W 24T
287.0
11M 1W 18T
140.1
11M 1W 43T
205.1
12M 0W 36T
112.6
11M 2W 13T
131.0
11M 2W 5T
159.2
12M 1W 2T
143.0
12M 2W 21T
160.8
12M 0W 29T
203.5
10M 2W 10T
250.3
10M 2W 28T
220.4
11M 0W 35T
134.7
11M 0W 22T
293.6
11M 0W 49T
159.1
11M 1W 3T
194.6
11M 2W 19T
271.9
11M 1W 11T
185.0
11M 0W 48T
240.6
11M 1W 28T
324.4
12M 0W 3T
260.2
12M 1W 36T
200.0
12M 0W 31T
178.2
12M 2W 0T
202.6
12M 2W 17T
217.5
12M 1W 43T
120.0
12M 2W 5T
132.3
01M 0W 32T
65.18
11M 0W 13T
164.2
11M 1W 40T
213.1
12M 2W 6T
211.3
01M 0W 33T
93.41
12M 2W 23T
284.7
02M 2W 14T
177.1
12M 0W 41T
191.6
12M 0W 15T
216.0
12M 1W 14T
142.5
12M 1W 20T
194.5
12M 1W 25T
231.2
12M 0W 40T
280.6
12M 0W 23T
179.7
12M 0W 9T
196.5
12M 0W 48T
308.4
12M 1W 7T
178.1
12M 1W 0T
160.0
12M 1W 30T
264.8
12M 1W 31T
123.9
12M 1W 32T
177.3
12M 1W 41T
260.4
12M 2W 27T
155.3
12M 2W 46T
122.8
12M 2W 20T
307.0
12M 2W 7T
170.9
12M 2W 11T
198.3
12M 2W 33T
133.9
12M 2W 2T
126.6
01M 0W 38T
40.48
01M 0W 37T
41.30
01M 0W 19T
108.6
01M 0W 39T
102.9
01M 0W 11T
102.4
01M 0W 16T
62.12
01M 0W 26T
72.61
01M 1W 3T
49.54
01M 1W 14T
74.30
01M 1W 23T
178.5
01M 1W 41T
115.6
01M 1W 32T
154.5
01M 0W 9T
112.1
02M 0W 4T
70.51
01M 0W 49T
148.1
01M 1W 45T
193.2
01M 2W 5T
229.9
01M 2W 40T
204.8
02M 0W 18T
193.3
02M 0W 33T
195.3
01M 0W 47T
58.68
01M 1W 49T
89.76
01M 0W 6T
138.8
01M 1W 18T
191.5
01M 1W 12T
160.7
01M 2W 4T
224.8
01M 2W 12T
191.8
02M 0W 6T
66.62
02M 0W 19T
169.7
02M 0W 22T
160.6
02M 1W 2T
114.1
02M 2W 22T
274.0
02M 1W 33T
219.0
03M 1W 1T
282.2
02M 2W 23T
217.4
03M 2W 24T
301.4
03M 0W 7T
194.3
02M 2W 38T
112.1
03M 0W 21T
163.8
03M 1W 17T
178.0
04M 0W 7T
104.7
03M 2W 41T
214.7
01M 2W 14T
155.8
01M 1W 31T
61.66
01M 2W 36T
140.8
01M 2W 26T
81.10
02M 0W 37T
143.3
01M 1W 36T
113.2
01M 1W 11T
104.5
01M 2W 41T
92.82
02M 0W 43T
73.85
01M 1W 17T
106.1
01M 2W 44T
139.2
01M 1W 0T
148.3
02M 0W 48T
141.5
02M 0W 42T
155.1
02M 1W 5T
149.2
02M 0W 1T
110.2
02M 2W 30T
128.9
02M 1W 31T
162.9
02M 2W 10T
186.0
02M 2W 16T
197.0
02M 0W 27T
55.77
02M 0W 46T
72.83
02M 1W 1T
160.9
02M 1W 6T
139.9
02M 1W 8T
86.81
02M 1W 27T
165.7
02M 2W 2T
168.7
02M 2W 35T
120.3
02M 0W 30T
105.1
03M 0W 0T
114.2
02M 1W 17T
144.7
02M 2W 7T
115.8
03M 0W 38T
132.6
03M 0W 19T
135.5
03M 1W 36T
160.1
02M 2W 37T
159.7
03M 1W 46T
159.9
04M 0W 36T
70.01
03M 1W 41T
158.9
04M 0W 10T
72.98
03M 0W 14T
139.6
03M 1W 28T
190.5
04M 0W 45T
81.85
03M 2W 15T
178.4
03M 2W 48T
148.6
02M 1W 43T
127.3
02M 1W 39T
160.9
02M 1W 7T
121.4
02M 1W 49T
146.8
02M 2W 9T
128.0
02M 2W 15T
123.3
02M 2W 40T
171.6
02M 2W 28T
141.3
02M 2W 29T
130.1
03M 1W 2T
156.3
03M 0W 26T
140.1
02M 1W 44T
163.6
03M 0W 30T
218.3
03M 2W 39T
127.6
04M 0W 18T
120.3
03M 1W 47T
199.9
03M 1W 39T
117.4
03M 2W 27T
198.7
03M 2W 32T
191.6
04M 0W 23T
85.85
04M 0W 29T
84.74
02M 1W 32T
143.6
02M 1W 11T
155.2
02M 2W 32T
156.0
02M 2W 34T
177.8
03M 0W 12T
155.4
03M 0W 2T
162.9
03M 2W 18T
201.8
03M 1W 8T
254.4
04M 0W 47T
69.58
02M 1W 46T
189.8
02M 2W 36T
154.0
02M 1W 28T
180.7
02M 2W 43T
176.8
02M 1W 38T
167.6
02M 1W 35T
226.6
02M 2W 47T
132.0
02M 2W 48T
185.8
03M 2W 8T
162.5
04M 0W 12T
83.44
03M 2W 30T
231.6
04M 0W 46T
98.31
11M 1W 16T
197.9
11M 2W 48T
156.0
12M 1W 48T
129.9
01M 0W 3T
51.24
01M 0W 28T
60.40
01M 0W 25T
51.30
09M 1W 25T
199.3
09M 1W 0T
180.5
08M 0W 24T
185.5
08M 1W 14T
255.7
11M 0W 27T
123.2
11M 0W 41T
153.3
11M 0W 5T
199.6
10M 0W 24T
521.2
10M 1W 40T
186.6
10M 1W 17T
239.3
10M 2W 34T
377.8
11M 1W 45T
164.0
11M 2W 21T
151.6
12M 0W 4T
141.9
12M 1W 16T
130.5
01M 1W 5T
78.32
01M 2W 47T
76.50
11M 0W 24T
177.0
11M 1W 33T
199.1
11M 2W 46T
167.5
12M 2W 8T
143.3
12M 0W 43T
215.4
12M 1W 0T
160.0
12M 1W 38T
236.5
12M 2W 46T
122.8
12M 2W 6T
211.3
11M 2W 31T
163.9
11M 2W 12T
305.4
11M 2W 29T
231.4
12M 0W 34T
255.3
12M 0W 16T
215.6
12M 1W 13T
241.2
02M 0W 26T
74.02
02M 1W 15T
145.5
02M 0W 0T
44.09
03M 0W 1T
201.3
03M 0W 6T
150.7
03M 0W 15T
127.6
07M 2W 2T
206.2
08M 2W 12T
152.2
09M 1W 43T
156.4
09M 2W 42T
155.5
07M 0W 40T
61.41
07M 0W 15T
55.89
07M 1W 26T
83.50
07M 1W 29T
92.94
07M 2W 27T
151.0
07M 0W 25T
139.3
08M 1W 8T
199.9
08M 2W 49T
233.1
10M 1W 10T
197.3
10M 2W 5T
277.3
11M 2W 4T
115.2
12M 1W 47T
112.1
12M 2W 14T
132.7
11M 2W 8T
187.8
11M 1W 34T
154.3
11M 1W 0T
201.6
11M 2W 0T
178.7
12M 0W 2T
133.5
12M 0W 45T
178.9
02M 1W 23T
113.2
02M 2W 20T
129.4
02M 1W 26T
245.6
02M 0W 38T
145.3
02M 2W 6T
259.8
02M 1W 20T
169.0
03M 0W 9T
266.7
03M 0W 5T
176.5
12M 1W 21T
156.6
12M 1W 11T
151.7
12M 2W 9T
121.6
01M 0W 23T
54.77
01M 1W 46T
108.9
01M 1W 33T
101.3
01M 2W 31T
120.7
01M 2W 32T
102.1
02M 0W 15T
80.65
02M 0W 47T
96.71
02M 1W 22T
132.0
02M 2W 31T
126.1
03M 1W 22T
173.9
03M 0W 18T
164.4
03M 0W 16T
149.4
07M 0W 6T
99.31
07M 1W 41T
118.4
07M 0W 23T
162.2
07M 1W 44T
145.6
07M 2W 13T
211.3
07M 1W 33T
158.5
08M 0W 17T
151.3
07M 2W 41T
272.5
08M 1W 10T
150.4
08M 2W 36T
191.2
09M 0W 0T
153.4
09M 1W 11T
231.2
07M 2W 15T
246.4
08M 0W 13T
137.7
08M 1W 40T
160.1
09M 0W 12T
188.7
10M 0W 19T
493.3
07M 0W 24T
82.32
07M 1W 45T
207.9
07M 2W 18T
204.9
07M 2W 28T
298.5
08M 0W 31T
132.9
07M 1W 49T
129.9
07M 1W 18T
106.3
07M 2W 16T
130.1
08M 0W 5T
64.80
08M 1W 19T
148.8
07M 2W 29T
265.3
08M 2W 42T
209.3
09M 0W 13T
197.9
09M 1W 39T
187.9
09M 2W 24T
221.6
09M 2W 19T
186.1
10M 2W 43T
245.1
11M 0W 12T
220.0
11M 0W 34T
173.4
11M 1W 5T
220.7
11M 2W 7T
215.5
11M 2W 26T
133.0
12M 0W 28T
193.4
12M 2W 28T
144.8
12M 1W 24T
193.7
12M 2W 4T
216.4
12M 2W 10T
212.1
01M 0W 27T
69.58
01M 1W 37T
95.19
01M 2W 21T
105.3
07M 1W 0T
161.9
11M 2W 1T
111.2
11M 1W 12T
198.1
11M 2W 34T
228.5
12M 0W 38T
139.6
08M 0W 3T
139.2
08M 1W 5T
139.8
07M 2W 23T
157.2
09M 1W 22T
143.6
08M 1W 13T
92.15
10M 1W 11T
140.7
11M 0W 32T
164.6
10M 1W 1T
185.9
10M 2W 2T
303.4
07M 2W 20T
249.4
08M 1W 16T
166.0
09M 0W 18T
137.9
09M 2W 11T
203.8
10M 2W 40T
164.8
10M 1W 39T
202.9
07M 1W 35T
118.2
08M 2W 21T
167.5
10M 0W 0T
287.8
09M 1W 42T
187.0
07M 2W 19T
181.2
08M 0W 42T
132.5
08M 0W 23T
147.4
08M 1W 43T
168.0
10M 0W 5T
361.6
10M 2W 27T
220.3
11M 2W 44T
143.1
11M 2W 14T
235.5
07M 0W 31T
93.05
08M 2W 30T
188.8
09M 2W 48T
184.2
09M 2W 7T
251.0
10M 1W 42T
186.6
11M 1W 4T
265.9
12M 0W 10T
187.3
07M 2W 32T
161.0
08M 0W 44T
111.0
08M 1W 48T
109.1
09M 1W 31T
176.8
10M 0W 15T
306.0
11M 0W 43T
192.6
12M 1W 40T
209.7
01M 2W 48T
165.2
01M 2W 39T
60.95
02M 2W 4T
132.7
09M 2W 34T
186.6
11M 0W 25T
173.0
12M 2W 23T
284.7
07M 0W 17T
60.64
07M 1W 40T
116.0
08M 1W 24T
128.4
08M 2W 18T
153.0
09M 0W 4T
143.6
09M 1W 15T
136.2
10M 0W 4T
254.8
10M 1W 0T
129.0
10M 2W 26T
161.7
11M 0W 7T
135.2
01M 0W 46T
156.4
07M 0W 42T
102.3
07M 1W 34T
170.9
08M 0W 4T
122.9
08M 2W 48T
171.3
08M 2W 14T
229.3
09M 2W 31T
210.9
09M 2W 15T
286.8
10M 0W 38T
354.9
11M 0W 28T
119.4
12M 0W 32T
121.0
07M 2W 1T
198.0
07M 2W 34T
192.2
08M 1W 32T
141.8
08M 1W 44T
201.9
09M 0W 25T
185.3
09M 1W 12T
222.7
10M 1W 31T
155.1
10M 2W 24T
167.4
11M 1W 15T
140.7
12M 2W 39T
112.0
07M 1W 28T
134.0
08M 0W 38T
157.8
08M 2W 40T
141.4
09M 2W 16T
154.6
10M 0W 36T
209.3
12M 0W 14T
106.9
07M 1W 6T
175.8
07M 2W 14T
236.9
08M 0W 29T
81.55
08M 1W 45T
110.5
09M 1W 21T
138.6
07M 0W 3T
125.3
07M 1W 27T
171.6
07M 2W 39T
147.1
08M 2W 46T
122.3
09M 0W 26T
133.4
10M 1W 12T
106.2
11M 0W 14T
133.5
01M 1W 2T
186.6
01M 0W 13T
41.88
02M 1W 18T
201.6
02M 2W 39T
201.9
07M 1W 25T
183.8
09M 0W 6T
159.3
10M 0W 2T
321.5
11M 1W 37T
330.5
12M 1W 45T
169.9
01M 1W 22T
56.44
02M 1W 0T
226.4
03M 0W 37T
253.4
09M 1W 36T
205.1
09M 2W 18T
356.0
11M 0W 16T
255.5
11M 2W 9T
159.5
12M 1W 22T
140.9
01M 0W 18T
56.21
02M 2W 49T
214.4
08M 1W 42T
210.7
08M 2W 10T
220.6
09M 1W 27T
250.0
11M 1W 23T
243.9
11M 1W 17T
162.1
12M 0W 37T
138.9
12M 1W 12T
181.6
02M 1W 34T
190.7
08M 0W 10T
161.1
08M 2W 6T
312.3
09M 0W 20T
261.8
09M 2W 1T
175.0
10M 0W 10T
289.3
10M 2W 11T
380.6
10M 1W 30T
161.7
10M 2W 49T
241.0
11M 1W 32T
162.3
07M 0W 20T
90.48
07M 2W 8T
273.8
08M 1W 27T
233.2
08M 2W 25T
295.3
09M 2W 14T
269.7
10M 0W 47T
287.4
09M 2W 28T
186.6
10M 0W 29T
270.7
10M 1W 5T
140.6
10M 2W 6T
223.8
12M 0W 6T
149.7
12M 2W 3T
144.4
08M 1W 4T
129.2
08M 2W 27T
177.4
09M 0W 22T
214.9
08M 2W 4T
209.9
09M 0W 38T
201.6
10M 1W 9T
152.8
10M 2W 19T
207.9
11M 2W 25T
156.9
12M 0W 17T
162.1
12M 2W 12T
166.6
07M 2W 36T
251.9
08M 1W 9T
159.9
08M 2W 1T
226.8
09M 2W 13T
172.8
10M 0W 43T
253.7
09M 0W 45T
155.5
10M 0W 13T
226.5
10M 2W 25T
204.8
07M 1W 2T
180.1
07M 2W 21T
234.8
08M 0W 8T
135.0
04M 0W 32T
63.06
03M 0W 35T
230.6
03M 1W 38T
268.1
07M 1W 22T
143.5
07M 1W 31T
225.6
07M 1W 14T
128.9
07M 1W 7T
147.2
07M 1W 30T
198.2
07M 0W 2T
83.10
07M 1W 48T
173.7
07M 2W 38T
190.3
07M 1W 47T
162.4
07M 2W 48T
211.8
07M 0W 43T
146.0
07M 2W 24T
338.7
07M 2W 31T
174.0
07M 2W 35T
178.8
07M 2W 40T
292.3
07M 2W 43T
186.3
07M 2W 44T
267.8
08M 0W 9T
118.2
08M 0W 11T
157.8
08M 0W 25T
193.1
08M 0W 32T
134.9
08M 0W 0T
120.9
08M 2W 19T
184.4
09M 0W 44T
184.3
08M 1W 39T
136.7
08M 2W 43T
192.3
09M 0W 34T
219.3
09M 1W 41T
137.8
09M 2W 44T
248.0
10M 0W 28T
333.6
10M 1W 13T
170.8
10M 2W 45T
162.8
09M 0W 48T
147.1
09M 1W 14T
179.8
10M 0W 48T
258.6
08M 0W 28T
116.0
11M 1W 47T
166.5
11M 2W 42T
204.4
12M 0W 22T
156.1
12M 1W 9T
140.8
12M 2W 42T
161.9
08M 2W 41T
192.4
09M 0W 10T
201.9
09M 2W 41T
252.2
09M 2W 35T
174.8
09M 1W 2T
215.2
10M 0W 27T
393.7
10M 1W 27T
132.3
09M 1W 10T
235.2
09M 2W 47T
274.3
10M 0W 34T
362.1
10M 0W 25T
400.8
11M 1W 39T
188.5
10M 1W 36T
206.6
08M 0W 20T
153.4
08M 0W 7T
121.9
08M 1W 25T
197.8
08M 0W 26T
186.5
08M 1W 26T
101.4
08M 1W 36T
239.4
09M 0W 43T
246.4
08M 2W 33T
277.5
09M 2W 46T
309.9
08M 0W 46T
113.7
08M 1W 38T
158.8
08M 1W 47T
134.9
08M 2W 0T
170.3
08M 2W 15T
204.5
10M 1W 20T
232.2
10M 2W 23T
274.6
11M 2W 19T
271.9
12M 0W 42T
233.2
12M 2W 47T
189.8
09M 0W 23T
146.5
08M 1W 34T
141.3
08M 1W 1T
200.8
08M 0W 6T
110.5
08M 2W 11T
186.5
08M 2W 45T
231.2
09M 0W 24T
123.3
09M 0W 41T
237.4
09M 2W 6T
248.2
10M 2W 10T
250.3
10M 0W 7T
359.8
11M 0W 22T
293.6
11M 1W 49T
273.2
12M 0W 21T
204.2
12M 1W 33T
201.5
12M 2W 16T
200.5
08M 1W 31T
126.6
09M 1W 49T
210.5
09M 2W 40T
281.5
10M 1W 47T
254.5
08M 2W 3T
154.6
09M 1W 24T
165.9
09M 0W 29T
224.4
10M 0W 22T
370.1
11M 2W 35T
228.7
11M 0W 21T
158.6
12M 2W 11T
198.3
01M 0W 11T
102.4
12M 0W 48T
308.4
12M 1W 41T
260.4
09M 0W 36T
197.3
09M 0W 37T
164.8
09M 1W 3T
255.7
09M 1W 18T
176.3
09M 2W 26T
244.3
10M 0W 37T
379.5
10M 1W 34T
239.0
10M 2W 41T
241.8
09M 0W 31T
164.4
09M 0W 3T
195.5
10M 0W 30T
350.6
10M 0W 17T
379.9
09M 2W 9T
214.8
10M 1W 44T
219.8
10M 1W 4T
162.1
10M 2W 36T
238.1
10M 2W 35T
185.3
11M 1W 29T
230.9
09M 1W 34T
177.7
09M 2W 0T
200.3
09M 0W 15T
184.4
09M 1W 40T
189.2
10M 1W 19T
209.1
10M 1W 2T
189.5
10M 2W 8T
220.7
09M 1W 6T
198.3
09M 2W 23T
254.6
10M 1W 16T
233.7
11M 1W 7T
208.1
11M 1W 35T
179.9
11M 2W 37T
179.4
12M 1W 39T
222.8
01M 1W 28T
91.68
01M 2W 46T
88.06
02M 2W 24T
124.0
03M 0W 28T
150.0
09M 2W 3T
174.8
09M 2W 10T
203.8
10M 0W 12T
292.4
10M 1W 26T
144.2
11M 0W 8T
137.5
11M 1W 43T
205.1
11M 2W 5T
159.2
12M 0W 49T
196.9
12M 1W 26T
142.9
12M 2W 15T
165.7
09M 2W 30T
227.8
09M 2W 32T
188.7
10M 0W 33T
239.3
10M 0W 21T
346.7
10M 0W 26T
212.7
10M 1W 28T
154.3
10M 1W 41T
121.2
10M 2W 12T
213.4
09M 2W 43T
216.4
10M 0W 14T
256.6
10M 2W 4T
183.0
10M 2W 28T
220.4
11M 0W 35T
134.7
10M 1W 45T
155.8
10M 2W 44T
227.4
11M 0W 4T
132.5
11M 0W 49T
159.1
11M 1W 3T
194.6
11M 0W 26T
124.0
11M 1W 1T
140.9
11M 0W 2T
119.1
03M 0W 35T
230.6
11M 1W 28T
324.4
11M 1W 22T
190.5
11M 2W 6T
164.8
11M 2W 11T
257.4
11M 1W 27T
201.7
12M 0W 46T
159.3
12M 1W 27T
156.8
11M 2W 17T
163.4
11M 1W 10T
268.9
11M 2W 33T
145.1
11M 2W 32T
236.6
12M 0W 25T
155.6
12M 0W 20T
226.8
12M 0W 3T
260.2
12M 0W 47T
211.0
12M 1W 44T
193.8
12M 1W 36T
200.0
12M 2W 24T
199.7
12M 2W 37T
231.9
12M 0W 31T
178.2
12M 2W 0T
202.6
12M 1W 34T
149.4
12M 1W 43T
120.0
12M 2W 5T
132.3
01M 0W 32T
65.18
01M 1W 7T
81.27
03M 1W 2T
156.3
02M 1W 14T
219.1
11M 0W 48T
240.6
11M 1W 11T
185.0
11M 2W 40T
159.4
12M 0W 30T
121.9
11M 1W 18T
140.1
11M 2W 13T
131.0
01M 1W 4T
132.1
01M 2W 33T
115.3
02M 0W 41T
105.1
03M 0W 4T
220.8
02M 0W 17T
83.46
10M 0W 3T
312.6
12M 1W 15T
162.7
12M 2W 17T
217.5
02M 0W 33T
195.3
02M 2W 14T
177.1
02M 1W 48T
170.6
09M 2W 33T
193.6
10M 0W 1T
334.5
10M 2W 38T
215.0
11M 2W 30T
211.0
12M 0W 29T
203.5
12M 0W 7T
215.2
12M 1W 49T
210.6
01M 2W 1T
115.0
09M 0W 16T
148.5
11M 1W 9T
291.3
11M 1W 6T
282.0
11M 2W 3T
260.1
12M 1W 37T
264.3
02M 2W 9T
128.0
12M 2W 49T
152.4
02M 1W 4T
196.9
10M 1W 37T
262.5
10M 2W 20T
291.7
11M 2W 24T
287.0
11M 0W 38T
230.9
11M 2W 22T
251.3
12M 0W 35T
227.4
12M 1W 5T
225.1
12M 2W 31T
242.1
01M 0W 33T
93.41
11M 2W 20T
160.5
02M 2W 13T
164.8
09M 1W 44T
292.8
09M 2W 45T
399.7
10M 1W 7T
227.8
10M 2W 7T
364.5
11M 0W 31T
243.0
11M 1W 26T
297.6
12M 0W 0T
248.8
01M 2W 40T
204.8
12M 2W 25T
160.8
02M 0W 29T
97.04
08M 2W 23T
139.6
09M 2W 27T
134.2
10M 0W 32T
492.1
11M 0W 0T
176.8
11M 0W 47T
145.7
11M 2W 49T
204.5
11M 1W 13T
275.7
11M 2W 43T
218.1
11M 0W 36T
217.2
09M 1W 28T
180.1
11M 1W 21T
231.4
10M 2W 30T
305.7
01M 1W 32T
154.5
11M 2W 23T
173.9
12M 0W 12T
188.7
12M 1W 35T
146.9
09M 1W 4T
231.2
10M 2W 22T
274.3
01M 0W 34T
61.04
02M 2W 17T
149.3
03M 0W 29T
210.0
07M 1W 21T
146.4
08M 1W 23T
127.3
08M 2W 2T
150.4
08M 2W 28T
256.3
09M 0W 30T
220.9
11M 0W 13T
164.2
11M 1W 40T
213.1
12M 2W 19T
175.6
08M 1W 0T
152.1
08M 1W 3T
105.6
08M 2W 22T
247.0
09M 2W 49T
276.3
10M 0W 6T
314.4
11M 0W 11T
151.4
12M 1W 2T
143.0
08M 1W 29T
180.8
07M 2W 46T
163.9
09M 1W 9T
241.1
08M 2W 7T
160.7
09M 0W 17T
161.4
10M 1W 32T
152.9
10M 2W 16T
240.5
11M 0W 15T
150.5
02M 1W 42T
170.4
03M 0W 17T
151.0
08M 2W 5T
255.6
08M 0W 12T
114.5
08M 1W 11T
155.4
09M 1W 45T
187.1
09M 2W 12T
208.1
10M 1W 6T
152.7
10M 2W 42T
178.7
12M 2W 21T
160.8
11M 2W 38T
142.2
09M 2W 20T
189.8
12M 0W 36T
112.6
02M 0W 9T
64.04
02M 1W 43T
127.3
01M 1W 29T
141.6
01M 2W 30T
116.3
02M 0W 23T
138.4
02M 1W 25T
204.5
02M 2W 18T
167.7
03M 0W 22T
194.9
12M 0W 18T
143.1
12M 0W 27T
112.5
12M 0W 41T
191.6
12M 0W 15T
216.0
12M 1W 10T
145.5
12M 1W 14T
142.5
12M 1W 20T
194.5
12M 1W 25T
231.2
12M 2W 34T
154.0
12M 0W 40T
280.6
12M 0W 23T
179.7
12M 0W 9T
196.5
12M 1W 7T
178.1
12M 2W 33T
133.9
12M 2W 2T
126.6
12M 1W 30T
264.8
12M 1W 31T
123.9
12M 1W 32T
177.3
12M 2W 27T
155.3
01M 0W 16T
62.12
01M 0W 26T
72.61
12M 2W 20T
307.0
01M 0W 9T
112.1
12M 2W 7T
170.9
01M 0W 38T
40.48
01M 1W 3T
49.54
01M 0W 37T
41.30
01M 0W 19T
108.6
01M 0W 39T
102.9
02M 0W 27T
55.77
01M 1W 14T
74.30
01M 1W 23T
178.5
01M 1W 41T
115.6
02M 1W 8T
86.81
02M 0W 4T
70.51
03M 0W 0T
114.2
01M 0W 49T
148.1
01M 1W 45T
193.2
01M 2W 5T
229.9
02M 0W 18T
193.3
01M 0W 47T
58.68
01M 1W 49T
89.76
01M 0W 6T
138.8
01M 1W 18T
191.5
01M 1W 12T
160.7
01M 2W 4T
224.8
01M 2W 12T
191.8
02M 0W 6T
66.62
02M 0W 19T
169.7
02M 0W 22T
160.6
02M 1W 2T
114.1
02M 2W 22T
274.0
02M 1W 33T
219.0
02M 2W 38T
112.1
03M 1W 1T
282.2
02M 2W 23T
217.4
03M 0W 21T
163.8
01M 2W 14T
155.8
01M 1W 31T
61.66
01M 2W 36T
140.8
02M 0W 37T
143.3
03M 0W 7T
194.3
01M 2W 26T
81.10
01M 1W 36T
113.2
01M 2W 41T
92.82
02M 0W 43T
73.85
02M 2W 16T
197.0
03M 0W 14T
139.6
01M 1W 11T
104.5
01M 1W 17T
106.1
01M 2W 44T
139.2
02M 0W 1T
110.2
02M 1W 31T
162.9
02M 2W 10T
186.0
01M 0W 24T
87.48
01M 1W 0T
148.3
02M 0W 30T
105.1
02M 0W 42T
155.1
02M 1W 6T
139.9
02M 2W 35T
120.3
02M 2W 37T
159.7
03M 0W 19T
135.5
02M 0W 48T
141.5
01M 1W 10T
78.12
02M 0W 46T
72.83
02M 1W 1T
160.9
02M 1W 5T
149.2
02M 1W 16T
133.7
02M 1W 27T
165.7
02M 2W 2T
168.7
02M 2W 30T
128.9
02M 1W 17T
144.7
02M 1W 44T
163.6
02M 1W 39T
160.9
02M 2W 7T
115.8
02M 2W 15T
123.3
02M 2W 40T
171.6
03M 0W 38T
132.6
03M 0W 26T
140.1
03M 0W 30T
218.3
02M 1W 7T
121.4
02M 2W 28T
141.3
03M 1W 32T
113.2
02M 1W 37T
207.7
03M 0W 45T
224.7
03M 2W 29T
143.0
04M 0W 41T
90.92
03M 2W 35T
173.9
07M 2W 22T
211.6
08M 2W 34T
198.2
08M 1W 2T
148.1
08M 2W 44T
230.2
09M 1W 5T
215.2
11M 1W 25T
182.4
02M 2W 5T
138.1
03M 0W 34T
140.4
03M 1W 45T
163.1
03M 2W 40T
176.8
03M 1W 6T
263.1
02M 2W 6T
259.8
03M 0W 9T
266.7
02M 2W 25T
209.4
03M 0W 27T
246.7
03M 1W 4T
200.4
03M 1W 10T
249.6
03M 2W 33T
238.3
03M 1W 13T
156.6
03M 1W 48T
247.6
03M 2W 28T
220.6
02M 2W 41T
113.7
03M 0W 23T
136.3
03M 0W 20T
206.5
03M 2W 3T
177.6
04M 0W 24T
83.73
03M 2W 38T
194.8
04M 0W 27T
58.27
03M 0W 4T
220.8
02M 1W 26T
245.6
02M 2W 20T
129.4
03M 1W 19T
277.4
03M 2W 23T
318.1
04M 0W 28T
122.4
12M 0W 27T
112.5
12M 1W 10T
145.5
02M 1W 14T
219.1
01M 1W 2T
186.6
02M 0W 38T
145.3
02M 1W 20T
169.0
03M 0W 5T
176.5
02M 1W 48T
170.6
02M 2W 24T
124.0
03M 0W 28T
150.0
12M 0W 18T
143.1
12M 2W 34T
154.0
03M 1W 35T
211.2
02M 0W 41T
105.1
01M 0W 46T
156.4
01M 2W 48T
165.2
01M 1W 28T
91.68
01M 2W 46T
88.06
02M 0W 17T
83.46
11M 2W 33T
145.1
12M 0W 25T
155.6
12M 2W 13T
160.4
01M 0W 45T
78.13
04M 0W 30T
83.01
12M 2W 25T
160.8
11M 2W 10T
155.8
12M 0W 11T
143.8
12M 1W 18T
140.6
12M 1W 44T
193.8
12M 2W 37T
231.9
03M 2W 4T
179.7
01M 2W 33T
115.3
01M 0W 34T
61.04
11M 1W 10T
268.9
11M 2W 32T
236.6
12M 0W 20T
226.8
01M 1W 29T
141.6
02M 0W 23T
138.4
02M 1W 25T
204.5
02M 2W 18T
167.7
03M 0W 22T
194.9
01M 1W 4T
132.1
12M 1W 13T
241.2
02M 1W 4T
196.9
11M 2W 17T
163.4
12M 1W 27T
156.8
12M 2W 24T
199.7
01M 2W 30T
116.3
11M 2W 12T
305.4
12M 0W 34T
255.3
02M 2W 13T
164.8
11M 1W 27T
201.7
12M 0W 46T
159.3
12M 0W 47T
211.0
01M 1W 7T
81.27
02M 0W 9T
64.04
11M 2W 29T
231.4
12M 0W 16T
215.6
03M 0W 29T
210.0
11M 1W 22T
190.5
11M 2W 6T
164.8
11M 2W 11T
257.4
12M 0W 42T
233.2
11M 2W 40T
159.4
12M 1W 15T
162.7
12M 1W 34T
149.4
12M 1W 26T
142.9
12M 2W 15T
165.7
11M 2W 30T
211.0
12M 0W 30T
121.9
12M 0W 49T
196.9
12M 2W 31T
242.1
12M 2W 39T
112.0
10M 0W 17T
379.9
10M 1W 28T
154.3
10M 2W 12T
213.4
10M 2W 38T
215.0
11M 1W 25T
182.4
11M 1W 6T
282.0
11M 2W 3T
260.1
12M 1W 37T
264.3
09M 2W 30T
227.8
10M 1W 26T
144.2
10M 2W 25T
204.8
11M 0W 32T
164.6
11M 0W 38T
230.9
11M 2W 22T
251.3
12M 0W 35T
227.4
12M 1W 5T
225.1
12M 1W 47T
112.1
09M 2W 10T
203.8
10M 0W 12T
292.4
10M 0W 13T
226.5
10M 2W 40T
164.8
10M 2W 7T
364.5
11M 0W 31T
243.0
11M 1W 26T
297.6
12M 0W 0T
248.8
10M 0W 3T
312.6
10M 1W 11T
140.7
10M 1W 16T
233.7
11M 1W 7T
208.1
11M 1W 35T
179.9
09M 2W 33T
193.6
10M 0W 0T
287.8
10M 0W 32T
492.1
11M 0W 0T
176.8
11M 0W 47T
145.7
12M 1W 39T
222.8
11M 2W 4T
115.2
12M 0W 14T
106.9
09M 1W 6T
198.3
09M 1W 40T
189.2
09M 2W 0T
200.3
10M 1W 4T
162.1
10M 2W 35T
185.3
11M 1W 21T
231.4
11M 2W 49T
204.5
11M 1W 15T
140.7
12M 2W 47T
189.8
12M 2W 16T
200.5
11M 2W 31T
163.9
12M 1W 45T
169.9
12M 1W 35T
146.9
12M 2W 42T
161.9
12M 2W 10T
212.1
08M 2W 27T
177.4
09M 0W 22T
214.9
09M 1W 27T
250.0
10M 1W 17T
239.3
10M 2W 11T
380.6
11M 0W 16T
255.5
12M 1W 38T
236.5
01M 2W 21T
105.3
08M 1W 4T
129.2
08M 2W 25T
295.3
09M 0W 20T
261.8
09M 2W 18T
356.0
10M 0W 24T
521.2
11M 1W 23T
243.9
12M 0W 43T
215.4
01M 0W 27T
69.58
08M 1W 14T
255.7
08M 2W 6T
312.3
10M 1W 47T
254.5
12M 0W 7T
215.2
12M 1W 49T
210.6
12M 2W 49T
152.4
01M 2W 1T
115.0
08M 0W 26T
186.5
08M 1W 27T
233.2
10M 0W 22T
370.1
11M 0W 21T
158.6
11M 2W 23T
173.9
11M 1W 39T
188.5
11M 2W 20T
160.5
11M 2W 26T
133.0
08M 0W 24T
185.5
08M 1W 26T
101.4
09M 0W 43T
246.4
09M 2W 46T
309.9
10M 0W 25T
400.8
11M 0W 11T
151.4
11M 1W 47T
166.5
11M 2W 42T
204.4
02M 0W 29T
97.04
03M 0W 17T
151.0
07M 0W 2T
83.10
07M 2W 43T
186.3
08M 0W 28T
116.0
09M 0W 48T
147.1
03M 2W 1T
170.3
07M 0W 12T
131.0
07M 1W 16T
197.1
07M 2W 5T
197.8
08M 0W 35T
86.74
08M 2W 26T
135.8
09M 0W 39T
110.4
09M 1W 33T
145.7
11M 0W 44T
161.3
11M 1W 30T
184.0
12M 1W 1T
174.6
12M 2W 45T
177.6
07M 0W 11T
74.81
07M 1W 12T
136.8
07M 2W 26T
138.5
08M 0W 19T
85.82
08M 1W 37T
120.5
02M 0W 39T
77.89
02M 2W 45T
174.4
07M 1W 0T
161.9
07M 2W 19T
181.2
08M 0W 42T
132.5
08M 2W 21T
167.5
03M 1W 44T
170.7
04M 0W 26T
63.00
07M 0W 22T
62.42
07M 1W 38T
103.9
10M 1W 49T
123.9
10M 2W 46T
177.9
11M 0W 30T
148.1
11M 2W 45T
166.5
12M 0W 39T
179.3
12M 2W 29T
237.6
02M 0W 16T
146.1
09M 0W 47T
136.8
11M 0W 23T
211.4
11M 2W 16T
229.3
12M 0W 8T
216.0
08M 2W 9T
182.6
09M 0W 8T
151.8
09M 1W 46T
161.0
10M 1W 8T
185.0
10M 2W 48T
367.8
10M 2W 17T
299.6
11M 0W 9T
176.9
11M 1W 19T
279.5
12M 0W 13T
227.9
08M 1W 6T
154.7
08M 2W 31T
187.9
09M 2W 22T
232.6
10M 0W 40T
352.1
10M 0W 39T
295.3
11M 2W 39T
227.3
01M 0W 48T
140.2
08M 1W 21T
148.1
08M 2W 16T
217.5
09M 0W 46T
212.7
10M 1W 23T
283.6
01M 1W 44T
85.36
07M 1W 11T
138.5
07M 2W 17T
237.1
08M 0W 45T
129.5
08M 2W 20T
233.3
09M 0W 14T
212.1
09M 1W 29T
229.5
09M 2W 25T
313.3
10M 0W 44T
464.8
01M 2W 13T
88.71
07M 0W 48T
82.50
07M 2W 45T
235.2
08M 0W 37T
114.5
08M 1W 12T
166.7
08M 2W 38T
304.9
09M 1W 37T
337.3
07M 1W 23T
124.6
07M 2W 47T
244.0
08M 0W 21T
151.6
08M 1W 35T
269.2
09M 0W 32T
316.7
09M 1W 8T
237.7
01M 0W 30T
85.10
12M 2W 4T
216.4
01M 1W 37T
95.19
02M 1W 42T
170.4
02M 2W 31T
126.1
07M 1W 35T
118.2
08M 0W 5T
64.80
12M 1W 24T
193.7
01M 1W 33T
101.3
02M 0W 47T
96.71
02M 1W 22T
132.0
07M 1W 18T
106.3
07M 2W 16T
130.1
11M 1W 12T
198.1
11M 2W 34T
228.5
12M 0W 28T
193.4
01M 0W 23T
54.77
01M 2W 32T
102.1
02M 0W 15T
80.65
07M 0W 24T
82.32
07M 1W 49T
129.9
07M 2W 29T
265.3
08M 0W 23T
147.4
10M 0W 5T
361.6
12M 0W 45T
178.9
12M 2W 28T
144.8
01M 2W 31T
120.7
07M 0W 23T
162.2
07M 1W 45T
207.9
07M 2W 28T
298.5
09M 1W 5T
215.2
09M 1W 39T
187.9
09M 2W 19T
186.1
10M 2W 43T
245.1
11M 1W 5T
220.7
04M 0W 22T
84.62
09M 0W 35T
234.5
12M 2W 38T
208.2
08M 1W 43T
168.0
08M 2W 42T
209.3
10M 1W 1T
185.9
10M 2W 2T
303.4
12M 1W 40T
209.7
07M 2W 18T
204.9
08M 0W 31T
132.9
09M 0W 13T
197.9
09M 1W 42T
187.0
09M 2W 24T
221.6
10M 1W 39T
202.9
01M 1W 46T
108.9
03M 2W 22T
176.0
04M 0W 42T
66.10
07M 2W 15T
246.4
08M 0W 13T
137.7
08M 1W 40T
160.1
09M 0W 12T
188.7
09M 1W 11T
231.2
11M 0W 12T
220.0
02M 1W 23T
113.2
03M 2W 45T
158.1
08M 2W 36T
191.2
11M 1W 0T
201.6
11M 2W 0T
178.7
01M 0W 13T
41.88
03M 0W 16T
149.4
07M 1W 44T
145.6
07M 1W 33T
158.5
07M 2W 41T
272.5
08M 1W 10T
150.4
11M 0W 25T
173.0
01M 1W 22T
56.44
03M 1W 37T
163.7
07M 0W 6T
99.31
07M 1W 41T
118.4
10M 2W 27T
220.3
01M 2W 39T
60.95
02M 0W 0T
44.09
03M 0W 15T
127.6
03M 1W 43T
138.3
07M 0W 31T
93.05
07M 2W 13T
211.3
08M 0W 17T
151.3
09M 0W 0T
153.4
10M 1W 42T
186.6
12M 2W 8T
143.3
01M 1W 5T
78.32
07M 0W 17T
60.64
07M 1W 40T
116.0
07M 2W 32T
161.0
08M 0W 44T
111.0
10M 0W 15T
306.0
10M 1W 0T
129.0
11M 2W 8T
187.8
12M 0W 4T
141.9
12M 1W 22T
140.9
01M 0W 18T
56.21
07M 0W 42T
102.3
07M 1W 34T
170.9
08M 1W 45T
110.5
08M 2W 40T
141.4
09M 2W 7T
251.0
10M 0W 4T
254.8
11M 0W 7T
135.2
11M 1W 34T
154.3
12M 2W 45T
177.6
07M 0W 15T
55.89
07M 1W 29T
92.94
07M 2W 39T
147.1
08M 0W 29T
81.55
09M 1W 31T
176.8
09M 2W 34T
186.6
10M 2W 26T
161.7
11M 2W 44T
143.1
04M 0W 16T
67.07
10M 2W 32T
353.4
02M 1W 32T
143.6
07M 1W 28T
134.0
07M 2W 34T
192.2
08M 1W 32T
141.8
08M 1W 44T
201.9
09M 2W 31T
210.9
11M 1W 4T
265.9
10M 2W 5T
277.3
02M 2W 4T
132.7
10M 1W 33T
258.7
02M 2W 32T
156.0
07M 0W 3T
125.3
07M 1W 6T
175.8
08M 1W 8T
199.9
08M 2W 49T
233.1
09M 0W 25T
185.3
10M 1W 10T
197.3
11M 0W 43T
192.6
11M 1W 45T
164.0
10M 2W 49T
241.0
03M 2W 11T
145.0
09M 2W 38T
362.8
03M 0W 12T
155.4
07M 0W 25T
139.3
07M 1W 27T
171.6
07M 2W 14T
236.9
08M 0W 38T
157.8
09M 1W 12T
222.7
09M 2W 15T
286.8
10M 1W 30T
161.7
10M 0W 38T
354.9
11M 0W 24T
177.0
11M 2W 48T
156.0
10M 1W 5T
140.6
10M 2W 6T
223.8
03M 0W 6T
150.7
10M 0W 20T
494.0
01M 2W 6T
100.5
07M 2W 27T
151.0
11M 0W 5T
199.6
11M 1W 33T
199.1
02M 1W 15T
145.5
09M 2W 5T
267.0
12M 2W 48T
235.5
07M 1W 26T
83.50
07M 2W 2T
206.2
10M 0W 2T
321.5
10M 1W 40T
186.6
10M 2W 34T
377.8
11M 0W 41T
153.3
04M 0W 15T
57.49
08M 2W 8T
261.1
12M 1W 46T
169.1
07M 0W 40T
61.41
07M 1W 25T
183.8
08M 2W 10T
220.6
09M 1W 36T
205.1
09M 1W 0T
180.5
09M 2W 28T
186.6
10M 0W 29T
270.7
02M 0W 26T
74.02
10M 1W 18T
233.7
12M 0W 24T
140.6
08M 0W 10T
161.1
08M 1W 42T
210.7
08M 2W 4T
209.9
09M 0W 38T
201.6
11M 1W 17T
162.1
02M 1W 18T
201.6
10M 2W 18T
297.2
11M 1W 46T
169.0
07M 0W 20T
90.48
07M 2W 8T
273.8
07M 2W 36T
251.9
08M 1W 9T
159.9
08M 2W 1T
226.8
10M 1W 9T
152.8
10M 2W 19T
207.9
02M 2W 39T
201.9
10M 0W 23T
395.3
01M 1W 42T
81.03
07M 0W 12T
131.0
07M 1W 16T
197.1
07M 2W 5T
197.8
11M 1W 32T
162.3
11M 2W 45T
166.5
10M 1W 49T
123.9
01M 2W 47T
76.50
02M 1W 0T
226.4
03M 1W 29T
239.6
07M 2W 49T
291.1
12M 2W 30T
161.5
07M 0W 11T
74.81
07M 1W 12T
136.8
11M 0W 27T
123.2
11M 1W 16T
197.9
11M 2W 15T
233.1
10M 1W 24T
167.2
01M 2W 10T
171.3
03M 0W 37T
253.4
03M 1W 3T
261.0
03M 2W 31T
301.3
10M 2W 15T
183.2
01M 0W 42T
61.69
07M 0W 22T
62.42
08M 0W 35T
86.74
09M 1W 33T
145.7
09M 2W 13T
172.8
10M 0W 43T
253.7
10M 2W 46T
177.9
11M 0W 44T
161.3
11M 1W 30T
184.0
02M 1W 49T
146.8
02M 1W 11T
155.2
02M 1W 46T
189.8
02M 2W 29T
130.1
02M 2W 34T
177.8
02M 2W 36T
154.0
03M 2W 34T
161.4
03M 0W 2T
162.9
02M 2W 33T
112.1
02M 1W 24T
171.1
03M 0W 46T
130.6
03M 0W 48T
184.6
03M 1W 40T
170.4
03M 1W 31T
228.3
03M 1W 22T
173.9
03M 2W 25T
188.0
02M 2W 1T
117.0
02M 2W 8T
109.5
03M 0W 31T
188.2
02M 2W 3T
140.1
03M 0W 49T
93.48
03M 1W 0T
132.1
03M 1W 11T
175.0
03M 1W 21T
185.9
03M 1W 5T
128.0
03M 1W 34T
208.4
03M 1W 2T
156.3
03M 2W 26T
155.3
03M 2W 39T
127.6
03M 1W 4T
200.4
03M 1W 32T
113.2
03M 1W 42T
219.4
03M 2W 2T
183.0
03M 2W 3T
177.6
03M 2W 17T
192.1
03M 2W 18T
201.8
03M 2W 40T
176.8
03M 2W 43T
134.0
03M 2W 47T
227.7
04M 0W 11T
100.7
04M 0W 24T
83.73
04M 0W 39T
82.15
04M 0W 47T
69.58
04M 0W 41T
90.92
W
#
#
A
A
A
03M 2W 5T
124.5
04M 0W 10T
72.98
01M 1W 4T
132.1
01M 0W 46T
156.4
01M 1W 28T
91.68
09M 2W 46T
309.9
10M 0W 21T
346.7
11M 0W 8T
137.5
11M 1W 43T
205.1
12M 0W 49T
196.9
01M 0W 33T
93.41
12M 0W 34T
255.3
01M 2W 46T
88.06
09M 2W 30T
227.8
10M 1W 26T
144.2
11M 2W 1T
111.2
02M 0W 41T
105.1
12M 2W 28T
144.8
09M 2W 10T
203.8
10M 0W 12T
292.4
10M 2W 38T
215.0
11M 1W 25T
182.4
12M 0W 38T
139.6
01M 2W 33T
115.3
12M 2W 23T
284.7
12M 2W 9T
121.6
09M 2W 3T
174.8
10M 0W 3T
312.6
10M 2W 25T
204.8
11M 0W 32T
164.6
11M 1W 6T
282.0
11M 2W 3T
260.1
12M 0W 35T
227.4
12M 2W 31T
242.1
12M 1W 13T
241.2
12M 2W 6T
211.3
12M 2W 42T
161.9
09M 2W 33T
193.6
10M 0W 0T
287.8
10M 1W 11T
140.7
10M 2W 40T
164.8
11M 0W 38T
230.9
11M 1W 26T
297.6
11M 2W 22T
251.3
12M 1W 5T
225.1
12M 0W 14T
106.9
11M 2W 29T
231.4
12M 0W 16T
215.6
12M 1W 38T
236.5
12M 1W 9T
140.8
09M 2W 11T
203.8
10M 1W 7T
227.8
10M 2W 7T
364.5
11M 0W 31T
243.0
12M 0W 0T
248.8
11M 2W 4T
115.2
11M 1W 37T
330.5
09M 1W 6T
198.3
09M 2W 2T
159.5
10M 0W 41T
273.6
10M 2W 29T
184.8
10M 1W 31T
155.1
10M 2W 24T
167.4
10M 1W 17T
239.3
10M 2W 11T
380.6
11M 0W 16T
255.5
11M 1W 23T
243.9
11M 2W 31T
163.9
12M 0W 43T
215.4
12M 2W 19T
175.6
04M 0W 38T
94.99
11M 0W 45T
123.1
11M 2W 28T
142.7
08M 1W 37T
120.5
08M 2W 26T
135.8
09M 0W 39T
110.4
09M 1W 7T
186.5
11M 0W 30T
148.1
12M 1W 29T
202.6
12M 2W 41T
255.8
01M 1W 38T
148.0
07M 1W 38T
103.9
07M 2W 26T
138.5
09M 0W 2T
166.2
09M 2W 36T
209.8
10M 0W 42T
301.0
10M 0W 9T
273.2
10M 2W 39T
249.7
03M 1W 23T
216.7
10M 2W 0T
305.9
10M 1W 38T
168.3
09M 1W 13T
162.6
10M 1W 43T
163.0
03M 2W 21T
221.9
11M 0W 40T
205.7
10M 1W 22T
196.9
08M 0W 20T
153.4
08M 1W 29T
180.8
08M 2W 22T
247.0
09M 1W 10T
235.2
09M 2W 47T
274.3
10M 0W 25T
400.8
12M 2W 10T
212.1
10M 2W 33T
225.6
12M 0W 39T
179.3
12M 2W 29T
237.6
02M 0W 16T
146.1
02M 2W 45T
174.4
08M 0W 43T
163.3
10M 0W 35T
305.0
11M 2W 16T
229.3
12M 0W 8T
216.0
12M 1W 42T
320.8
01M 0W 48T
140.2
09M 0W 47T
136.8
10M 0W 18T
243.6
10M 2W 17T
299.6
08M 0W 19T
85.82
08M 1W 6T
154.7
08M 2W 9T
182.6
09M 0W 8T
151.8
09M 1W 46T
161.0
09M 2W 29T
201.0
10M 1W 8T
185.0
10M 2W 48T
367.8
03M 0W 42T
198.9
08M 1W 28T
196.5
09M 0W 33T
165.3
10M 0W 39T
295.3
12M 0W 26T
296.0
12M 1W 3T
214.1
12M 2W 38T
208.2
08M 0W 1T
120.7
12M 1W 18T
140.6
07M 2W 0T
175.6
08M 0W 40T
139.9
10M 0W 40T
352.1
11M 0W 29T
223.7
11M 1W 14T
257.4
12M 2W 48T
235.5
07M 0W 29T
52.55
07M 1W 11T
138.5
07M 2W 9T
174.3
08M 0W 33T
114.3
08M 1W 21T
148.1
08M 2W 31T
187.9
09M 2W 22T
232.6
09M 2W 21T
224.5
10M 2W 32T
353.4
11M 0W 46T
205.9
12M 0W 24T
140.6
08M 1W 7T
159.2
12M 2W 13T
160.4
07M 0W 48T
82.50
07M 1W 23T
124.6
07M 2W 17T
237.1
08M 0W 45T
129.5
09M 1W 16T
241.8
10M 1W 23T
283.6
10M 2W 18T
297.2
11M 1W 46T
169.0
08M 2W 39T
136.8
01M 0W 45T
78.13
07M 0W 13T
95.21
07M 1W 46T
135.1
07M 2W 45T
235.2
08M 0W 37T
114.5
08M 2W 16T
217.5
09M 0W 46T
212.7
09M 1W 29T
229.5
09M 2W 25T
313.3
10M 0W 44T
464.8
10M 1W 33T
258.7
10M 1W 18T
233.7
11M 0W 45T
123.1
09M 1W 48T
172.4
10M 1W 35T
134.4
09M 1W 2T
215.2
01M 1W 37T
95.19
07M 2W 47T
244.0
08M 2W 20T
233.3
09M 0W 14T
212.1
09M 2W 38T
362.8
10M 0W 20T
494.0
10M 2W 15T
183.2
09M 2W 17T
187.6
10M 0W 46T
207.7
11M 2W 10T
155.8
09M 0W 10T
201.9
01M 0W 27T
69.58
08M 0W 21T
151.6
08M 1W 12T
166.7
08M 2W 38T
304.9
09M 0W 32T
316.7
09M 2W 5T
267.0
10M 0W 23T
395.3
10M 1W 38T
168.3
10M 2W 0T
305.9
11M 0W 40T
205.7
07M 0W 38T
75.02
07M 1W 5T
102.9
07M 2W 30T
184.9
09M 2W 2T
159.5
10M 0W 41T
273.6
07M 0W 10T
112.0
07M 1W 36T
188.0
08M 1W 35T
269.2
08M 2W 8T
261.1
09M 1W 8T
237.7
07M 2W 25T
312.0
08M 0W 49T
210.2
08M 1W 28T
196.5
09M 0W 35T
234.5
07M 0W 4T
149.1
07M 1W 24T
226.3
07M 2W 49T
291.1
08M 0W 43T
163.3
10M 0W 35T
305.0
10M 1W 22T
196.9
01M 1W 44T
85.36
10M 1W 29T
135.9
08M 0W 12T
114.5
10M 0W 6T
314.4
11M 0W 11T
151.4
11M 1W 39T
188.5
11M 2W 20T
160.5
12M 0W 12T
188.7
12M 1W 35T
146.9
03M 2W 8T
162.5
01M 2W 13T
88.71
10M 2W 29T
184.8
08M 2W 28T
256.3
08M 2W 7T
160.7
10M 1W 32T
152.9
10M 2W 16T
240.5
11M 1W 0T
201.6
04M 0W 12T
83.44
02M 0W 39T
77.89
09M 1W 26T
149.5
08M 1W 0T
152.1
08M 1W 11T
155.4
09M 0W 17T
161.4
09M 1W 45T
187.1
09M 2W 12T
208.1
11M 2W 0T
178.7
12M 1W 11T
151.7
02M 2W 24T
124.0
02M 1W 34T
190.7
07M 1W 4T
113.6
08M 0W 28T
116.0
09M 2W 7T
251.0
11M 0W 25T
173.0
12M 0W 45T
178.9
03M 0W 28T
150.0
01M 0W 30T
85.10
08M 1W 41T
138.8
08M 0W 25T
193.1
08M 2W 10T
220.6
09M 1W 31T
176.8
10M 1W 42T
186.6
10M 2W 27T
220.3
01M 2W 1T
115.0
07M 1W 9T
170.1
08M 2W 29T
151.6
07M 2W 43T
186.3
08M 0W 10T
161.1
08M 1W 42T
210.7
08M 2W 48T
171.3
10M 0W 15T
306.0
02M 1W 4T
196.9
04M 0W 17T
119.5
07M 2W 10T
184.2
07M 2W 38T
190.3
09M 0W 4T
143.6
09M 2W 34T
186.6
04M 0W 22T
84.62
07M 2W 4T
307.2
08M 0W 14T
102.0
07M 2W 1T
198.0
09M 1W 15T
136.2
12M 0W 2T
133.5
12M 1W 21T
156.6
02M 2W 13T
164.8
08M 1W 7T
159.2
07M 0W 21T
86.41
07M 2W 48T
211.8
08M 0W 9T
118.2
08M 1W 32T
141.8
08M 2W 18T
153.0
10M 0W 4T
254.8
11M 0W 7T
135.2
11M 1W 34T
154.3
11M 2W 44T
143.1
03M 0W 29T
210.0
12M 1W 46T
169.1
11M 0W 42T
140.3
07M 2W 8T
273.8
08M 0W 4T
122.9
08M 1W 24T
128.4
10M 1W 0T
129.0
10M 2W 26T
161.7
03M 1W 48T
247.6
01M 1W 42T
81.03
09M 0W 40T
154.6
07M 1W 48T
173.7
07M 2W 44T
267.8
10M 1W 6T
152.7
11M 0W 15T
150.5
11M 1W 47T
166.5
11M 2W 42T
204.4
12M 0W 22T
156.1
03M 2W 28T
220.6
12M 2W 30T
161.5
07M 1W 1T
130.4
07M 1W 33T
158.5
07M 2W 41T
272.5
09M 2W 20T
189.8
02M 0W 29T
97.04
01M 0W 42T
61.69
07M 2W 7T
229.5
07M 1W 30T
198.2
07M 2W 40T
292.3
09M 1W 14T
179.8
10M 0W 48T
258.6
10M 2W 42T
178.7
02M 0W 0T
44.09
11M 2W 28T
142.7
07M 0W 34T
78.26
07M 1W 7T
147.2
07M 2W 35T
178.8
08M 0W 0T
120.9
08M 1W 39T
136.7
08M 2W 43T
192.3
09M 0W 48T
147.1
01M 1W 22T
56.44
09M 1W 48T
172.4
10M 1W 35T
134.4
07M 1W 14T
128.9
07M 2W 31T
174.0
08M 1W 2T
148.1
09M 1W 5T
215.2
01M 2W 39T
60.95
04M 0W 38T
94.99
09M 2W 17T
187.6
10M 0W 46T
207.7
11M 0W 42T
140.3
07M 1W 31T
225.6
07M 2W 24T
338.7
08M 0W 8T
135.0
08M 2W 44T
230.2
09M 0W 34T
219.3
01M 1W 44T
85.36
09M 0W 33T
165.3
09M 2W 2T
159.5
10M 0W 41T
273.6
10M 1W 29T
135.9
10M 2W 29T
184.8
07M 1W 22T
143.5
07M 2W 22T
211.6
08M 1W 5T
139.8
01M 2W 13T
88.71
08M 2W 39T
136.8
09M 1W 26T
149.5
07M 1W 2T
180.1
07M 2W 21T
234.8
08M 0W 3T
139.2
03M 1W 23T
216.7
08M 0W 1T
120.7
08M 1W 41T
138.8
08M 2W 29T
151.6
07M 1W 0T
161.9
07M 2W 20T
249.4
08M 1W 16T
166.0
08M 2W 21T
167.5
03M 2W 21T
221.9
07M 0W 38T
75.02
08M 0W 14T
102.0
09M 0W 40T
154.6
07M 1W 35T
118.2
07M 2W 19T
181.2
08M 0W 42T
132.5
11M 2W 26T
133.0
02M 0W 39T
77.89
02M 2W 45T
174.4
07M 1W 1T
130.4
07M 1W 18T
106.3
07M 2W 16T
130.1
10M 2W 2T
303.4
11M 0W 12T
220.0
11M 2W 7T
215.5
03M 0W 42T
198.9
07M 2W 7T
229.5
07M 0W 24T
82.32
07M 1W 49T
129.9
07M 2W 29T
265.3
08M 0W 23T
147.4
10M 1W 39T
202.9
11M 1W 5T
220.7
11M 2W 14T
235.5
02M 1W 34T
190.7
04M 0W 17T
119.5
07M 0W 34T
78.26
07M 0W 23T
162.2
07M 1W 45T
207.9
07M 2W 28T
298.5
09M 1W 39T
187.9
09M 2W 19T
186.1
10M 0W 5T
361.6
10M 1W 1T
185.9
10M 2W 43T
245.1
11M 0W 34T
173.4
11M 1W 4T
265.9
12M 1W 22T
140.9
01M 0W 25T
51.30
07M 2W 18T
204.9
08M 0W 31T
132.9
08M 1W 43T
168.0
08M 2W 36T
191.2
10M 2W 5T
277.3
11M 0W 43T
192.6
11M 2W 8T
187.8
12M 0W 37T
138.9
12M 2W 3T
144.4
07M 0W 42T
102.3
07M 1W 29T
92.94
07M 2W 39T
147.1
08M 1W 44T
201.9
09M 1W 12T
222.7
09M 2W 15T
286.8
10M 1W 10T
197.3
11M 0W 24T
177.0
11M 1W 33T
199.1
11M 2W 25T
156.9
01M 0W 28T
60.40
07M 0W 15T
55.89
07M 1W 28T
134.0
07M 2W 34T
192.2
08M 1W 8T
199.9
08M 2W 49T
233.1
10M 0W 38T
354.9
11M 0W 5T
199.6
11M 1W 16T
197.9
11M 2W 9T
159.5
12M 2W 29T
237.6
07M 0W 3T
125.3
07M 1W 6T
175.8
07M 2W 14T
236.9
08M 0W 38T
157.8
10M 1W 30T
161.7
10M 2W 49T
241.0
11M 0W 41T
153.3
11M 1W 17T
162.1
01M 0W 48T
140.2
07M 1W 27T
171.6
07M 2W 27T
151.0
09M 0W 25T
185.3
10M 1W 5T
140.6
10M 2W 6T
223.8
11M 0W 44T
161.3
11M 1W 32T
162.3
11M 2W 48T
156.0
12M 2W 38T
208.2
07M 0W 25T
139.3
07M 1W 26T
83.50
07M 2W 2T
206.2
09M 1W 0T
180.5
09M 2W 28T
186.6
10M 0W 29T
270.7
10M 1W 49T
123.9
10M 2W 46T
177.9
11M 1W 30T
184.0
11M 2W 45T
166.5
01M 0W 30T
85.10
07M 0W 40T
61.41
07M 1W 25T
183.8
08M 2W 4T
209.9
09M 0W 38T
201.6
10M 1W 9T
152.8
10M 2W 19T
207.9
11M 0W 27T
123.2
07M 0W 20T
90.48
07M 2W 36T
251.9
08M 1W 9T
159.9
09M 1W 33T
145.7
09M 2W 13T
172.8
10M 0W 43T
253.7
10M 1W 24T
167.2
11M 0W 30T
148.1
07M 0W 12T
131.0
07M 1W 16T
197.1
07M 2W 5T
197.8
08M 0W 35T
86.74
08M 2W 1T
226.8
09M 0W 39T
110.4
09M 1W 7T
186.5
07M 0W 11T
74.81
07M 1W 12T
136.8
08M 0W 19T
85.82
08M 1W 37T
120.5
08M 2W 26T
135.8
09M 0W 2T
166.2
10M 0W 42T
301.0
10M 2W 33T
225.6
11M 1W 20T
220.0
11M 2W 15T
233.1
12M 0W 39T
179.3
12M 1W 6T
210.9
07M 0W 22T
62.42
07M 2W 26T
138.5
08M 1W 6T
154.7
08M 2W 9T
182.6
10M 2W 39T
249.7
11M 0W 23T
211.4
11M 2W 16T
229.3
09M 2W 36T
209.8
10M 0W 18T
243.6
10M 1W 43T
163.0
10M 2W 48T
367.8
11M 0W 9T
176.9
11M 1W 41T
214.2
11M 2W 27T
205.5
07M 1W 38T
103.9
07M 2W 0T
175.6
08M 0W 40T
139.9
09M 1W 13T
162.6
09M 2W 29T
201.0
10M 1W 8T
185.0
10M 2W 13T
225.7
07M 0W 29T
52.55
07M 1W 11T
138.5
08M 0W 45T
129.5
09M 0W 47T
136.8
09M 1W 46T
161.0
09M 2W 22T
232.6
10M 0W 40T
352.1
10M 1W 3T
192.1
10M 2W 1T
348.6
07M 2W 17T
237.1
09M 0W 8T
151.8
09M 1W 29T
229.5
09M 2W 21T
224.5
10M 0W 39T
295.3
10M 1W 23T
283.6
11M 0W 19T
257.8
11M 1W 42T
362.0
11M 2W 47T
341.4
08M 2W 31T
187.9
09M 0W 14T
212.1
09M 1W 16T
241.8
10M 1W 33T
258.7
11M 1W 14T
257.4
11M 2W 39T
227.3
08M 1W 21T
148.1
08M 2W 20T
233.3
09M 0W 46T
212.7
09M 2W 25T
313.3
10M 0W 44T
464.8
10M 1W 18T
233.7
10M 2W 18T
297.2
11M 0W 46T
205.9
08M 0W 33T
114.3
08M 1W 12T
166.7
08M 2W 16T
217.5
09M 0W 7T
289.7
09M 1W 37T
337.3
09M 2W 38T
362.8
10M 0W 23T
395.3
07M 0W 13T
95.21
07M 1W 36T
188.0
07M 2W 47T
244.0
08M 1W 35T
269.2
08M 2W 38T
304.9
09M 0W 32T
316.7
09M 2W 5T
267.0
10M 0W 46T
207.7
07M 2W 25T
312.0
08M 0W 49T
210.2
08M 1W 28T
196.5
08M 2W 39T
136.8
09M 0W 33T
165.3
07M 0W 4T
149.1
07M 1W 24T
226.3
07M 2W 3T
332.0
08M 0W 43T
163.3
08M 1W 7T
159.2
08M 2W 29T
151.6
09M 0W 40T
154.6
04M 0W 5T
88.05
03M 0W 40T
123.6
03M 1W 25T
124.2
03M 1W 7T
136.9
03M 1W 33T
201.0
03M 2W 6T
137.2
03M 2W 7T
145.0
03M 2W 19T
124.4
w
N
C
C
D
W
03M 1W 14T
194.6
03M 0W 11T
139.1
03M 1W 34T
208.4
03M 1W 32T
113.2
03M 1W 42T
219.4
03M 0W 13T
134.7
03M 1W 12T
260.8
03M 2W 2T
183.0
03M 2W 17T
192.1
03M 2W 9T
183.9
03M 2W 16T
172.0
03M 2W 26T
155.3
03M 2W 43T
134.0
03M 2W 47T
227.7
03M 2W 49T
194.1
04M 0W 9T
119.8
04M 0W 11T
100.7
04M 0W 39T
82.15
04M 0W 49T
83.59
03M 2W 5T
124.5
W
#
#
A
A
A
w
N
C
C
D
W
W
#
#
A
A
A
w
N
C
C
D
W
W
#
#
A
A
A
w
N
C
C
D
W
W
#
#
A
A
A
w
N
C
C
D
W
W
#
#
A
A
A
w
N
C
C
D
W
02M 1W 28T
F g 3 S x sets o top c cha ns constructed w th s d ng w ndows o s zes one to s x For
each set o top c cha ns every top c cha n starts at the same vert ca pos t on W th n
each top c cha n top cs are tempora y ordered the o dest (first) top c at the top and
go ng down to the most recent top c at the bottom The number o nodes nd cates
the tota number o top cs that are connected w th one or more s m ar top cs B ue
nodes are those that be ong to the argest top c cha n n the ast set o top c cha ns
constructed w th the s d ng w ndow o s ze s x B ue nodes start out n the first set
o top c cha ns as sma top c cha ns and they agg omerate as the s ze o s d ng w ndow ncreases The u -s ze figure s ava ab e at http //u ab ka st ac kr/research/top ccha n/
Sliding Window Size The s ze of the s d ng w ndow s a so an mportant
factor for construct ng the top c cha ns If we use a s d ng w ndow of s ze one
t means that we on y cons der the prev ous t me s ce to find the s m ar top cs
for the top cs of the current t me s ce However th s Markov assumpt on s not
genera y he pfu as s m ar top cs can appear over non-consecut ve t me s ces
so proper cons derat on of the s d ng t me w ndow s needed to capture these
top c trends
We vary the s ze of the s d ng w ndow from one to s x and observe the
changes n the resu t ng top c cha ns F gure 3 shows the top c cha ns of s ze
greater than one for the var ous w ndow s zes w th the r descr pt ve stat st cs
F rst the number of nodes nd cates the tota number of top cs that be ong n
top c cha ns Th s number exc udes s ng eton top cs and shows how many top cs
v” ú™ョ・ û3
m
.
¹ケ
8
7ュ タ*
. ú™ョ・ O=ニ
V
û3 タ* )· µ
~。 。é
v
d@ µ* タ* û
O
”. チm
ョ
ソ4 タ* û3 )
ャ
V
!
̨ʼ
ッ Ñ}K
アフ OP 3イ ö
ú
ùP
#
レィ öÊレィ sR fý
アフ 3イ
|
| V"
z5
。é O=ニ
3ã
5 ノs K
アフ ™ィ タm
Ó| タ? ヤ̀z ョË
ö
ィ タV
ÛI mィ Zc fý
08M 2P 25T
company, business, market, manufacturing, production
J チm íù4 z
$
oK アí
:
Chian, Europe, automobile, Russia, EU, FTA,
5
zョ
R- ョケ4 ヲc ョケ
hÏ
î
ケ
automobile, Europe, Hyundai Motors, KIA Motors, German
i Motors, customer, oil
。 6|4 l i
ミÈ 4
│̶ s
タ* û3 d@ µ
î ¥( ョË
mケ z?4
d@ タ*
ミ4 3X Ë34 ョî P
íù
customer, representative, Toyota, Hyundai Motors, Japan
Fig. 4. Detecting focus shifts using difference of a word probability along the topic
chain. Each rectangle represents a topic node, and contains top probability words.
Edges connect two similar topics within a sliding window of size six, and the words next
to the edge are the named entities whose probabilities are changed the most between
two topics. xxM yyP zzT represent month, period, and topic number, respectively.
Ö
^7
V l. Ÿs Aù @
:ö
OP 'Ô ùP O,
Jí U£& ùヤ ・ョ) = Ë
3 z& *̅
・
)· タ*
fý X ? ゚ミ4 íù
゚ミ íù4
$
|e
'Ô ùP OP O, ïG
fý タ?
V? O, 'Ô
µ Pユ ニ„
J }ù
$
$
J O=ニ
z
zシ:P ョO
5(
.
3 Ôョ
*m
3
J Zc Ÿs
OP ùP 'Ô O,
6.2
ミ@
Jí タ? シ? (|@イ|4
.
ィ
@ 3î|
・
Focus Shifts
When we construct topic chains, we find that there are long topic chains and
short or singleton topic chains. Long topic chains tend to cover very general
topics such as politics, economy, and sports, and we call them long-term topics.
4̅ *̅ 'Ô
v” }v lf
タ* ソ4 )· û3 d@
& U£& Zc ゚ミ4 &m ゚ミ
=ケ =ケ4 @
öP 4
>* =4= ィ< íù 。é öÊレィ s
4Ï
4̅ *̅ |4 'Ô :ö
out of 1,400 total, are matched with one or more similar topics within the time
window. The number of nodes increases at a faster rate from window size one
to four and at a slower rate from window size four to six, and through that, we
can see that similar topics do not necessarily appear in consecutive time slices.
Other graph characteristics also change with the size of the sliding window as
shown in Figure 3. The total number of topic chains decreases as we increase the
window size. This means many of the distributed small topic chains merge as the
size of the sliding window increases. This is further evidenced by increases of both
the average chain size and the average chain depth. The width of topic chain is
the maximum number of topics of the same time slice in that chain. Unlike other
increasing characteristics, the average width of the topic chain remains stable
throughout the size of the sliding window. This is expected because topics of the
same time slice represent different aspects of mainstream news.
Figure 3 illustrates how similar topics agglomerate as we vary the size of the
sliding window. We painted in blue nodes of the largest topic chain at a sliding
window of size six. We also painted the same nodes at the other sizes of the
sliding window. At the window size of one, there are fourteen separate topic
chains painted in blue. These chains join together to form larger chains as we
increase the size of the sliding window.
'Ô ùP Ôョ ž4
Šj4 3O| û
シ@ タ*
)· タ* û3 ソ4 d@
4̅ |4 'Ô *Ì ̅
。é =4= アí V" エŽ öィ ィ<
ú
'Ô 4̅ |4 エ™ uタ
|4 <= 4̅ öP 'Ô
3
ョ
uタ メu 'Ô
ミ 3O| V l.
Samsung Electronics, LG, Europe, Samsung, mobile phone
0
.
v
öP 4̅ *̅ 'Ô ンツ
z? íù 3î
09M 0P 22T
market, sales, automobile, product, company
oil, solar, patent, energy, Samsung
ソ4 タ* û3 )
Šj4 Xî タ?
Oil, carbon dioxide, Posco, natural ga
07M 2P 36T
technology, develop, environment, energy, produce
08M 2P 27T
automobile, production, energy, sales, market
09M 0P 20T
company, industry, technology, field, recruitmen
, Chian, business
hÏ ̲
battery, KIA Motors, Hyundai Motors, Electronic car, gasoline
solar, automobile, Toyota, hydrogen, GM Daewoo
%
green, Busan, robot, solar, employee
=ケ jù íù z? íù4 ョX V l. zョ
automobile, hybrid, Hyundai Motors, Japan, T
ùP öP 'Ô |4 s
08M 1P 4T
automobile, vehicle, electric power, electricity,model
Samsung electornics, LG, patent, Samsung, China
│
íù ̇ë
$
タ* û3 ソ4
07M 0P 12T
green, industry, develop, technology, fi
07M 1P 16T
develop, technology, automobile, investment, industry
Posco, Qualcomm, black, shipbuilding, california
08M 1P 14T
company, business, corporation, firm, technology
V l. ィ<
ö
。 5O h*ù
z5 h
"
green, solar, Japan, energy, carbon
ョî ョX| $H ョX û
@
07M 0P 11T
automobile, Vietnam, KIA Motors, vehicle, sales
T
.
ョË
)· タ* ソ4 û3 d@
Jí V l. 6|4 V" X? Zc
Interpreting Long-Term Topic Chains Looking at a long-term chain is like
looking at a section of the newspaper. Many of the long-term topic chains could
be labelled as “politics”, “business”, or “sports”, and the topics in those chains
reflect a wide variety of subjects within those general news categories. There
are also long-term topics, such as H1N1, which are more specific news items but
last for a long time. Our topic chains contain more helpful information for interpreting these long-term topics. For example, you can look at the “H1N1” topic
chain and read off when the topic first emerged and when it disappeared. You
can also see that the topic evolved from talking about “swine flu”, to “travel”,
to “vaccinations” and “deaths”.
Named Entities in Topic Evolution Looking at the topic chains, where each
node is shown with the top probability words for each individual topic, we can
see the general evolution of the topic chain, but it is difficult to interpret the
evolution to see what happened. This is because the words that represent the
individual topics may be too general and occur in many topics throughout the
topic chain. For example, words like season, home run, game, and coach are
always top probability words in a topic chain about baseball. Those frequently
occurring top words tell us what the general topic trend is, but it may be more
interesting to see how the focus shifts for each topic within the chain.
To identify the words that can help to understand the focus of the topic chain
at each time slice, we hypothesized that the words tagged as named entities–
people, places and organizations–would be good discriminating words of the
different focuses within the topic chain. We illustrate these named entities with
the most changes in probabilities in Figure 4. Each rectangle represents a topic
with the top five probability words. An edge connects two similar topics, and the
words next to the edge are the named entities that change the most between the
two topics. For example, topics 1 and 3 are both about the automobile industry,
but the named entities green, solar, Japan, and energy, show that the focus is
on green energy for topic 3. We can indeed find a related news article from
the time period of topic 3 with the headline “Toyota makes eco-friendly solar
car”. Also we can see the evolution of the topic from 2 to 3. Topic 2 represents
the general green (environmental-friendly) industry. By incorporating the focus
words automobile, hybrid, Hyndai Motors, and Toyota this topic evolves into
the topic of environmental-friendly automobiles, topic 3. From topic 3 to 4, the
electric car and its battery problems received attention from news, and from 4
to 5, other alternative sources of energy, solar and hydrogen became the focus.
6.3
Short Topic Chains and Singleton Topics
We discussed long topic chains in the previous section, but short topic chains–
chains of two or three topics, or singleton topics–are important for two reasons.
First, most of the short topic chains are about temporary issues. If a topic
lasts over a long period of time, it would become part of a long topic chain. That
means singleton topics and short topic chains are likely to be about temporary
Table 2. Examples of single node topic chains. First to sixth topics are temporary
issues. First issue refers to the missile launch from North Korea, second issue is related
to the death of Michael Jackson, fifth issue is related to the romance of Korean top
actor and actress, and sixth issue is talking about Arbor Day on April 5. These are
typical cases of temporary issues. However, the last example is not a coherent topic.
0P
0P
0P
2P
0P
0P
0P
Date
07M 2009Y
07M 2009Y
10M 2009Y
12M 2009Y
01M 2010Y
04M 2010Y
02M 2010Y
Topic
North Korea, missile, launch, range, UN Security Council, ship, navy, East sea, ballistic missile
Jackson, family, funeral, cherish the memory of, Michael Jackson, son, LA, publish, report, death
melamine, dry milk, region, environment, investigation, food, pollution, mercury, produce
flight, airport, passenger, airplane, search, terror, time, security, explosion, aircraft
Hyesoo Kim, actor, 2010, ski, Haejin Ryu, once, 4, soul, colleague, lover
tree, recover, park, culture, movement, development, environment, ecology, forest, designation
Obama, Republicans, Jeju island, game, Jeju, golf, White house, Woods, gamers, budget
issues, and we can see that is true for the examples of singleton topics and short
chains listed in Table 2. Topics such as the death of Michael Jackson, reinforcing
airport security at the end of the year, and romance between top actors do not
last for a long time and can be considered as temporary issues.
Second, some of the singleton topics are useless. When we extract topics
with LDA, the results do not consist of only meaningful topics. Sometimes LDA
extracts topics that are not understandable as coherent topics. For example, the
last topic in Table 2 is not a coherent topic. Constructing topic chains leaves
bad results of LDA to be isolated as singleton topics. Conversely, topics that
form long topic chains tend to be coherent. Evaluation of topics found by LDA
is an on-going challenging research problem[17], so our topic chain framework
may offer one solution of evaluating topics found in a sequential corpus. We will
explore this in future work.
7
Discussions
In this paper, we proposed a framework for analyzing a corpus of news articles
over a contiguous time period. Our framework discovers topics from the corpus,
constructs topic chains using a topic similarity metric, identifies long-term topics and temporary issues, and detects focus shifts within each topic chain. An
important contribution in this work is a comparison of various topic similarity
metrics. We looked at six commonly used metrics and compared them using the
negative log likelihood of corpus.
A secondary use of the topic chains is as an analysis tool to evaluate the
quality of topics by a topic model. Most of the work on probabilistic topic modeling typically assume that the latent space is semantically meaningful, and so
the topics are not systematically evaluated. In this work, we found that most of
the topics that belong to long topic chains are semantically meaningful, whereas
singleton topics are less coherent. Further analysis of the relationship among
topics in the sequential corpus may find an effective way to analyze semantic
meaningfulness of the topics.
References
1. Allan, J.: Introduction to topic detection and tracking. Topic Detection and
Tracking, Event-based Information Organization (2002) 1–16
2. Blei, D., Lafferty, J.: Dynamic topic models. Proceedings of the 23rd International
Conference on Machine Learning (2006)
3. Wang, X., McCallum, A.: Topics over time: a non-markov continuous-time model
of topical trends. Proceedings of the 12th ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining (2006)
4. Blei, D., Ng, A., Jordan, M.: Latent dirichlet allocation. The Journal of Machine
Learning Research (2003) 993–1022
5. Nallapati, R., Feng, A., Peng, F., Allan, J.: Event threading within news topics. In:
Proceedings of the thirteenth ACM international conference on Information and
knowledge management. (2004)
6. Feng, A., Allan, J.: Finding and linking incidents in news. Proceedings of the
sixteenth ACM Conference on Information and Knowledge Management (2007)
7. Allan, J., Gupta, R., Khandelwal, V.: Temporal summaries of new topics. Proceedings of the 24th Annual International Conference on ACM SIGIR Research
and Development in Information Retrieval (2001)
8. Blei, D., Lafferty, J.: Topic models. Text Mining: Theory and Applications (2009)
71–93
9. Wang, X., Zhang, K., Jin, X., Shen, D.: Mining common topics from multiple asynchronous text streams. Proceedings of the Second ACM International Conference
on Web Search and Data Mining (2009)
10. Bolelli, L., Ertekin, Ş., Giles, C.: Topic and trend detection in text collections
using latent dirichlet allocation. Advances in Information Retrieval (2009)
11. Leskovec, J., Backstrom, L., Kleinberg, J.: Meme-tracking and the dynamics of
the news cycle. Proceedings of the 15th ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining (2009)
12. Shahaf, D., Guestrin, C.: Connecting the dots between news articles. Proceedings
of the 16th ACM SIGKDD International Conference on Knowledge Discovery and
Data Mining (2010)
13. Mei, Q., Zhai, C.: Discovering evolutionary theme patterns from text: an exploration of temporal text mining. Proceedings of the eleventh ACM SIGKDD
international conference on Knowledge discovery and Data Mining (2005)
14. Mei, Q., Liu, C., Su, H., Zhai, C.: A probabilistic approach to spatiotemporal theme
pattern mining on weblogs. Proceedings of the 15th International Conference on
World Wide Web (2006)
15. He, Q., Chen, B., Pei, J., Qiu, B., Mitra, P.: Detecting topic evolution in scientific
literature: how can citations help? Proceeding of the 18th ACM Conference on
Information and Knowledge Management (2009)
16. Newman, D., Asuncion, A., Smyth, P.: Distributed algorithms for topic models.
The Journal of Machine Learning Research 10 (2009) 1801–1828
17. Chang, J., Boyd-graber, J., Gerrish, S., Wang, C., Blei, D.M.: Reading tea leaves:
How humans interpret topic models. Advances in Neural Information Processing
Systems (2010)