GB2475293A - Analyser for ranking of items on list of items - Google Patents

Abstract

The analyser comprises one or more lists, each list having one or more items (e.g. hotels, famous people), and each item having a position on the list and characterised by one or more features. It has a first calculator which measures an assessment made of the features and position of each item by the users, which forms a user:item pair for each item assessed and the calculator determines an assessment for each of said user:item pairs. It has a second calculator which measures the activity of each of the users and determines a user rank for each of them. The analyser calculates an individual rank value for each item:user pair by modifying the assessment of the user:item pairs according to the user rank of each user and according to how recently an assessment of each item was made by the user. The analyser calculates an overall rank value for each item by summing the individual rank values, and the items are ordered within the lists according to their overall rank. The analyser allows users to assess collectively or collaboratively a large number of items.

Description

Ranking Analyser

Field of Invention

This invention reiates generaiiy to ranking of items on iists of items by users. More specifically, the invention is directed towards allowing users of a website to establish a collective ranking of items on a list of items.

Background of the Invention

U.S. Patent Appl. Pub. No. 2002/0194607 provides a method and system for periodically deriving an optimal batch broadcast schedule based on client demand feedback data from a distributed set of broadcast clients. The broadcast system includes an operation centre that broadcasts meta-data to a plurality of client systems. The meta-data describes a plurality of pieces of content that are in consideration for upcoming broadcasts by the server. Each client receives the broadcasted meta-data from and sends back a set of client demand feedback data to the operations centre, wherein the user feedback data reflects a ctientTs interest level in at least a portion of the pieces of content. The feedback data, which typically may include ratings and/or relative rankings, may be user-generated, automatically-generated, or a combination of the two. The system can then send a batch of content based on an aggregation of the feedback data in combination with available broadcast bandwidth and broadcast schedule window.

U.S. Patent Appl. Pub. No. 2005/0021390 provides a method of rating an item which includes receiving an item selected by a user, receiving a first profile associated with a first criteria, receiving a second profile associated w±th a second criteria, and rating the item based on the first profile and the second profile. Rating the item may include retrieving values associated w±th the item. The values include values associated with the first criteria and values associated with the second criteria. Rating the item may also include applying the first profile to the values associated with the first profile and applying the second profile to the values associated with the second profile. The method may also include receiving from a user an indication of how to interpret a value and evaluating the value positively or negatively based on the indication.

U.S. Patent Appl. Pub. No. 2008/0189272 provides methods and systems for collaborative ranking of a set of digital items such as videos in an online content hosting website. A set of digital content items is provided to users who wish to participate in the collaborative ranking of the set of digital content items. Content providers control which digital content items are represented ±n the set. The set of digital content can be in the form of a playlist. Alternatively, a playlist can be defined by one or more of the users of the website, or a playlist can be automatically generated based on available information from the website such as the top ten highest rated digital content items or the top ten most viewed digital content items. In one embodiment, a web server retrieves the set of digital content items from a digital content database to provide to valid users participating in the collaborative ranking of the set of digital content items. Each user is displayed the current ranking of each digital content item in the set and provides ranking feedback for each digital content item representing a vote that associates the digital content item with a particular ranking position.

A scoring module tallies all valid votes received from all valid users and calculates a new ranking score for each digital content item in the set. The new ranking scored is used to update the collaborative ranking of the predefined set of digital content.

These approaches to ranking items in a list typically allow the user to make a simple relative judgment on the position of an item in a list: for example, to move it up, move it down, or leave it in its current position. Another approach is to allow the user to make a subjective appraisal on an item based on a number of characteristics, for example, according to a star system. A further approach is to allow the user to contribute a free format review on an item. All assume the opinion of each of the users is of equal weight.

Disclosure of Invention

From the foregoing, it may be appreciated that a need has arisen for a an approach to ranking items on a list that treats the opinions of users according to a weighting system that gives each of the users a weight.

The analyser of the present invention comprises one or more lists, each list having one or more items, and each item having a position on the list and characterised by one or more features. It has a first calculator which measures an assessment made of the features and position of each item by the users, which forms a user:item pair for each item assessed and the calculator determines an assessment for each of said user:item pairs. It has a second calculator which measures the activity of each of the users and determines a user rank for each of them. The analyser calculates an individual rank value for each item:user pair by modifying the assessment of the user:item pairs according to the user rank of each user and according to how recently an assessment of of each item was made by the user. The analyser calculates an overall rank value for each item by summing the individual rank values, and said the items are ordered within the lists according to their overall rank values.

The ranking analyser is fully configucable on both a machine level and list-by-list. Relative weighting of the meters, and the profiles of components (such as time decay settings and assessment criteria) can he individuaiiy fine-tuned and taiiored to any subject matter.

Features assessed inciude the reiative position of the item in the iist (too high/too iow/fairiy ranked), charactecistics of the individuai items (each given an evaluation rating) and a free-format review of the item by users.

Brief Description of Drawings

For a more compiete expianation of the present invention and the technicai advantages thereof, reference is now made to the foiiowing description and the accompanying drawing in which: Figures 1-3 show features of the anaiyser of the present invention; Figure 4 shows a top-ievei page deiivered by a server; Figure 5 shows a mid-ievei page deiivered by a server; and Figure 6 shows a iow-ievei page deiivered by a server.

Best Mode for Carrying Out the Invent ion

Embodiments of the present invention and their technicai advantages may be better understood by referring to Figures 1-3.

The Ranking Analyser of the present invention allows a large number of Users (102) to assess, collectively or collaboratively, a large number of Items (104).

Users are individuals that have direct access to the Ranking Analyser.

Preferably the Users are registered Users.

Users may submit Items to the Ranking Analyser. Users may organise Items in lists (106) and further group these lists into Zones (108) Items are digital representations of any discrete activity which may be assessed in comparison with peers. Foc example, Items may be hotels, or they may be famous people.

The Ranking Analyser determines a position of an Item in a list of Items according to an assessment of one or more features of each Item made by one or more Users. A User can assess Items in a number of ways. For example: Stance -a User makes a judgement on an Item's position in a list relative to its neighbours. If the User thinks an Item is too low, he/she votes UP; if he/she thinks it is too high, he/she votes DOWN; and if he/she thinks the Item is fairly valued he votes STAY.

Rating -a User assesses the Item against a range of characteristics, assigning a score for each one.

Review -a User writes a free format review of an Item.

This assessment aotivity is represented by the arrows in Figure 1.

The Ranking Anaiyser inoiudes a time deoay meter (306) . More reoent assessments of an Item are iikeiy to more reievant and vaiuabie than those that are years, weeks or even hours oid. Aooordingiy, the time deoay meter measures how reoentiy an assessment of one or more features of eaoh Item was made by the one or more Users and the Ranking Anaiyser aooords a higher weight of opinion to more reoent assessments than oider ones and fade the weight of opinion over time. An assessment made by a User wouid thus beoome iess signifioant over time. A User oan restore the strength of their assessment of an Item by, for exampie, refreshing their Stanoe or by updating their Rating or Review. Note that beoause eaoh User oan oniy have a singie stanoe on eaoh Item refreshing a Stanoe simpiy restores its fuii vaiue and no more; this prevents vote fraud where Users attempt to manipuiate rankings artifioiaiiy with muitipie votes on an Item.

The Ranking Analyser inoludes a User Ranks Caloulator (202) . If the User has assessed an Item by Rating it, the weight given to a Stanoe is enhanoed.

Similarly if the User has assessed an Item by adding a Review, this inoreases the weight given to both Stanre and Rating values. In addition as Users gain experienoe, their opinion gains weight. This means that Users who provide the greatest oontributions have the greatest weight of opinion. Aroordingly, the User Ranks Caloulator measures the aotivity of eaoh of the one or more users.

The User Ranks Caloulator measures the effort a User puts into both oolleotive and oollaborative assessment (204, and see below), and/or a variety of other related funotions, suoh as oontent generation (adding/editing lists, Items), Networking (forming oontaot groups) and so on, ooupled with faotors suob as Credits (other Users assessing one's Items, adding Items to one's lists, eto) and freguenoy of visits. This data is oollated and used to determine a User Rank for eaoh User aooording to their experienoe. Furthermore the experienoe meter assigns experienoe weightings to all faotors, oompares them against thresholds, and oaloulates a User Ranking for eaoh User for both individual Zones and the entire Ranking Analyser. The more artivity and Credits rerorded, the higher the resulting User Rank. A User having a higher User Rank will have a greater influenre over Item rankings than a User having a lower User Rank. Having separate Zone ranks means that Users ran develop isolated expert ratings in areas of their speoifio interest. The data for eaoh User for both individual Zones and the entire Ranking Analyser ran be used to produoe a User Rank (206) User Ranks ran be used to rontrol the soope of artivities that any User ran perform. Whereas a novire User may make a simple Stanre assessment on an Item, it is preferable to restrirt ronseguential artions, surh as rhanging the strurture of a well-established list to experienred Users. Individual List configurations can aiso be refined to dictate permissions. The Activity permissions nodule (208) then determines what each user may do.

The Ranking algorithm (302) collates all individual assessments by Users on Items. It applies User Rank weightings (206) . It applies specific List settings (304) . It applies the effects of time decay (306) using a calculation cycle. It also applies system wide configuration settings (308), that allow the administrator to dial up or down the overall importance of Ranks, Time decay etc. Using all inputs, the ranking algorithm calculates an individual Rank Value for each Item:User pair. Items are then given an overall Rank Value, which is the sum of all their individual Rank Values. Finally Items are ordered (310) within Lists: the Item with the highest overall Rank Value appears at position number 1. The output of the canking algorithm is a set of continuously updated, dynamic Lists, on any subjects defined by the Users. As the ordering of Lists constantly change, Users may reassess Items based on their new positions, perhaps judging an Item that was formerly too low now to be too high.

The foregoing describes how the Ranking Analyser can establish a collective opinion built up by many Users making individual assessments of many Items. A further feature of the Ranking Analyser is its ability to allow Users to work collaboratively to build an assessment framework and to refine a range of criteria, for any List of Items, the Ranking Analyser allows Users to define: attribute categories -these are facts relating to the List (which also work as List refinement filters) and are different for different Lists, e.g. attribute categories for wine and for golf courses will be very different; ranking behaviour -Lists have a natural volatility, for example News Stories will change more quickly than Hotels, and time decay profiles can be configured at a List level for Stances, Ratings, and Reviews, adjusting both the importance and speed of opinion decay, or switching it off completely.

Collaborative assessment relates to all activities involved in individual assessment of Items and in creating a coherent logical framework for assessments to be made. The high degree of configurability in the Ranking Analyser means that discrete groups in the community at large can work together to assess anything that can be represented digitally.

The previously described method for calculating user ranks is suitable for an open environment, with large numbers of users contributing to many subject matters. However in other circumstances, User Ranks calculation can be configured differently, or simply assigned by an Administrator. This would be appropriate in any closed environment where weight of opinion might be allocated according to criteria such as authority. For example: hiring managers, assessing job candidates, manager and peer performance reviews, teachers and students evaluating courses, administrative assessment of policies.

Typically the Lists (106) of Items (104) are displayed on a web site accessible by one or more Users (102) . It will be understood that the term Uweb site" represents any computer system adapted to serve content using any internetworking protocols, and is not intended to be limited to content uploaded or downloaded via the Internet or the HTTP protocol. In general, functions described in one embodiment as being performed on the server side can also be performed on the client side in other embodiments if appropriate.

In addition, the functionality attributed to a particular component can be performed by different or multiple components operating together. Information can be accessed through a client executing a browser connected to a web server via a network, which is typically the internet, but can also be any network, including but not limited to any combination of a LAN, a MAN, a WAN, a mobile, wired or wireless network, a private network, or a virtual private network. The client may include a variety of different computing devices.

Examples of client devices are personal computers, digital assistants, personal digital assistants, cellular phones, smart phones, mobile phones, or laptop computers. As will be obvious to one of ordinary skill in the art, the present invention is not limited to the devices listed above. The browser can include any application that allows users of client to access web pages on the World Wide Web. Suitable applications are Microsoft Internet Explorer, Google Chrome, Mozilla Firefox, Apple Safari or any application adapted to allow access to web pages on the World Wide Web.

List information to be ranked may be organised in any suitable manner, either flat or hierarchical. In one embodiment, Items are assembled into Lists of similar Items (e.g. amusing adverts), and the Lists are further assembled into Zones of similar Lists (e.g. Entertainment), and the Zones may be displayed at the highest level. This kind of arrangement is shown in Figures 4 to 6 and discussed in more detail below.

A first page (see Figure 4) shows a user-selectable list of the available Zones in a first panel (402) A second page (see Figure 5) shows a user-selectable list of the available lists within a zone in a first panel (402) A third page (see Figure 6) shows a user-selectable list of the available items within a list in a first panel (402) For both pages, a seoond panel (404) shows Faots pertinent to the Zone, list or Item seleoted, a third panel (406) shows Ratings information about how users have assessed various pertinent oharaoteristios, a fourth panel (408) shows Reviews users have made and a fifth panel (410) is a forum panel.

When a user seleots a Zone, List or Ztem in panel (402), the server oauses relevant Faots to be displayed in panel (404), Ratings to be displayed in panel (406), Reviews in panel (408), and Forum details in panel (410).

On any of the pages, a user may take a Stanoe about the position of a Zone, List or Ztem displayed in the list in panel (402) by olioking on a button or ioon (412) to signify a simple relative judgment on the position of a Zone, List or Item in the list in panel (402) . The button or ioon (412) has an upper part to signify that the user has a Ztanoe that the position should be higher, an upper part to signify that the position should be lower, and a middle part to signify that the position is oorreot.

On any of the pages, a user may give a Rating on an item based on a number of oharaoteristios displayed in panel (406) by olioking on a star rating seleotor (414) to signify a subjeotive appraisal on an item based on the oharaoteristios shown.

On any of the pages, a user may give a Review of an item in panel (408)