CN112612957B - Recommendation method of interest points and training method and device of recommendation model of interest points - Google Patents
Recommendation method of interest points and training method and device of recommendation model of interest points Download PDFInfo
- Publication number
- CN112612957B CN112612957B CN202011554561.8A CN202011554561A CN112612957B CN 112612957 B CN112612957 B CN 112612957B CN 202011554561 A CN202011554561 A CN 202011554561A CN 112612957 B CN112612957 B CN 112612957B
- Authority
- CN
- China
- Prior art keywords
- interest
- points
- candidate
- historical
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012549 training Methods 0.000 title claims abstract description 33
- 230000006399 behavior Effects 0.000 claims abstract description 41
- 238000012163 sequencing technique Methods 0.000 claims description 40
- 238000002372 labelling Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 238000004590 computer program Methods 0.000 abstract description 11
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000004891 communication Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000005065 mining Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 235000017337 Persicaria hydropiper Nutrition 0.000 description 2
- 240000000275 Persicaria hydropiper Species 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 235000011201 Ginkgo Nutrition 0.000 description 1
- 235000008100 Ginkgo biloba Nutrition 0.000 description 1
- 244000194101 Ginkgo biloba Species 0.000 description 1
- 241000722921 Tulipa gesneriana Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000004080 punching Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a recommendation method of interest points, a training method, a training device, training equipment, a storage medium and a computer program product of an interest point recommendation model, and relates to the fields of big data, artificial intelligence, deep learning and the like. The specific implementation scheme is as follows: determining historical interest points visited by the current user according to the historical behaviors of the current user; determining a plurality of candidate interest points according to the historical interest points; sorting the candidate interest points by using the labels of the candidate interest points; and recommending a plurality of candidate interest points according to the ranking. And constructing the interests of the current user by using the historical behaviors of the current user, and expanding the interests of the current user to obtain candidate interest points. And obtaining a recommendation sequence according to the labels of the candidate interest points, so that thousands of people and thousands of faces can be realized in the map application program by combining the interests of the user, the personalized requirements of the user are met, and the use experience of the user is improved.
Description
Technical Field
The application relates to the technical field of computers, in particular to the fields of big data, artificial intelligence, deep learning and the like.
Background
The usage habit of users for map application programs is mostly to search by self or find out the surrounding entries. When a user retrieves and discovers the surrounding results, the map generally returns a fixed result that is ranked based on the heat and quality of the points of interest, and the user experience is poor.
Disclosure of Invention
The application provides a recommendation method of an interest point, a training method, a training device, training equipment, a storage medium and a computer program product of an interest point recommendation model.
According to an aspect of the present application, there is provided a recommendation method for a point of interest, which may include the steps of:
determining historical interest points visited by the current user according to the historical behaviors of the current user;
determining a plurality of candidate interest points according to the historical interest points;
sorting the candidate interest points by using the labels of the candidate interest points;
and recommending a plurality of candidate interest points according to the ranking.
According to another aspect of the present application, there is provided a training method of a point of interest recommendation model, which may include the steps of:
for a plurality of interest point samples, obtaining labels and sequencing truth values of each interest point sample;
determining the weight of the label of the interest point sample;
obtaining a sequencing predicted value of each interest point sample according to the label of each interest point sample and the weight of the label by the interest point recommendation model to be trained;
training the interest point recommendation model to be trained according to the sequencing predicted value and the sequencing true value until the errors of the sequencing predicted value and the sequencing true value are within the allowable range.
According to a third aspect of the present application, there is provided a recommendation device for a point of interest, the device may include:
the historical interest point determining module is used for determining historical interest points visited by the current user according to the historical behaviors of the current user;
the candidate interest point determining module is used for determining a plurality of candidate interest points according to the historical interest points;
the sorting module is used for sorting the plurality of candidate interest points by using the label of each candidate interest point;
and the recommending module is used for recommending a plurality of candidate interest points according to the ordering.
According to a fourth aspect of the present application, there is provided a training apparatus for a point of interest recommendation model, the apparatus may include:
the interest point sample information acquisition module is used for acquiring labels and sequencing truth values of each interest point sample for a plurality of interest point samples;
the weight determining module is used for determining the weight of the label of the interest point sample;
the sequencing predicted value determining module is used for enabling the interest point recommendation model to be trained to obtain the sequencing predicted value of each interest point sample according to the label of each interest point sample and the weight of the label;
and the training module is used for training the interest point recommendation model to be trained according to the sequencing predicted value and the sequencing true value until the error of the sequencing predicted value and the sequencing true value is within the allowable range.
In a fifth aspect, an embodiment of the present application provides an electronic device, including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods provided by any one of the embodiments of the present application.
In a sixth aspect, embodiments of the present application provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method provided by any one of the embodiments of the present application.
According to another aspect of the application there is provided a computer program product comprising computer instructions which, when executed by a processor, implement the method of any of the embodiments of the application.
According to the technology disclosed by the application, the interests of the current user are constructed by utilizing the historical behaviors of the current user, and the interests are expanded to obtain candidate interest points. And obtaining a recommendation sequence according to the labels of the candidate interest points, so that thousands of people and thousands of faces can be realized in the map application program by combining the interests of the user, the personalized requirements of the user are met, and the use experience of the user is improved.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the application or to delineate the scope of the application. Other features of the present application will become apparent from the description that follows.
Drawings
The drawings are included to provide a better understanding of the present application and are not to be construed as limiting the application. Wherein:
FIG. 1 is a flow chart of a recommendation method for points of interest according to the present application;
FIG. 2 is a flow chart for ranking a plurality of candidate points of interest in accordance with the present application;
FIG. 3 is a flow chart of determining a plurality of candidate points of interest in accordance with the present application;
FIG. 4 is a flow chart of determining a plurality of candidate points of interest in accordance with the present application;
FIG. 5 is a flow chart of a training method of the point of interest recommendation model according to the present application;
FIG. 6 is a schematic diagram of a recommender in accordance with the present application;
FIG. 7 is a schematic diagram of a training apparatus for a point of interest recommendation model according to the present application;
FIG. 8 is a scene graph in which a recommendation method for points of interest may be implemented;
FIG. 9 is a block diagram of an electronic device for implementing a point of interest recommendation method and/or a training method for a point of interest recommendation model in accordance with an embodiment of the present application.
Description of the embodiments
Exemplary embodiments of the present application will now be described with reference to the accompanying drawings, in which various details of the embodiments of the present application are included to facilitate understanding, and are to be considered merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
As shown in fig. 1, in one embodiment, the present application relates to a method for recommending points of interest, which may include the following steps:
s101: determining historical interest points visited by the current user according to the historical behaviors of the current user;
s102: determining a plurality of candidate interest points according to the historical interest points;
s103: sorting the candidate interest points by using the labels of the candidate interest points;
s104: and recommending a plurality of candidate interest points according to the ranking.
The above-described scheme of the present application can be applied to map-type applications. The historical behavior of the current user may include search records, navigation records, etc. of the current user. Such as scenic spots searched by the current user, restaurants navigated, etc., can be used as the current user's historical behavior. In addition, the search record may also be a search for a sagged class. For example, search for a nearby chaffy dish store, a nearby bookstore, etc. The hotpot and bookstore can then be considered as different drop categories.
Under the condition that the current user searches or navigates, candidate interest points with higher correlation degree with the historical interest points can be obtained according to the historical interest points searched or navigated by the current user. The candidate points of interest may be determined based on geographic location, based on whether the candidate points of interest are the same vertical, based on average consumption, etc. Or, weights can be set for the interest points determined in different determining modes, and a predetermined number of interest points are screened out as candidate interest points in a weighted sum mode.
Each candidate point of interest may include at least one tag. For example, the label may be the number of times the point of interest is accessed as a candidate point of interest, the number of times the point of interest is accessed, and so on. And comprehensively comparing each candidate interest point by using the labels, and sequencing each interest point according to the comparison result.
For example, the ranking of each point of interest may be determined as a ratio of the number of times the point of interest is accessed to the number of times that the point of interest is candidate. Or, ordering only the number of times the point of interest is accessed, etc.
By ordering each candidate point of interest, the ordering may be utilized to recommend points of interest. For example, in the case where a current user search for "food" instruction is received. The "guess you like" module is added in the search list. The content of the method is that a plurality of candidate restaurants which are finished to be ranked are determined according to the historical behaviors of the current user. For another example, in the event that a current user click "view surrounding" instruction is received, the candidate points of interest may be presented in different manners. First, points of interest with the same characteristics can be displayed in an aggregate manner, such as 'ten around chafing dish first', 'around most popular net red punching card place', and the like. And secondly, the traditional recall result can be displayed after being reordered.
Through the scheme, the interests of the current user are built by using the historical behaviors of the current user, and the interests are expanded to obtain candidate interest points. And obtaining a recommendation sequence according to the labels of the candidate interest points, so that thousands of people and thousands of faces can be realized in the map application program by combining the interests of the user, the personalized requirements of the user are met, and the use experience of the user is improved.
Referring to fig. 2, in one embodiment, the ranking the candidate points of interest using the label of each candidate point of interest in step S102 includes:
s201: extracting a plurality of labels of each candidate interest point;
s202: determining the weight of each tag of each candidate interest point;
s203: and for each candidate interest point, inputting each label of the candidate interest point and the weight of each label into a pre-trained sequencing model to obtain a sequencing result of the candidate interest point.
The label of candidate points of interest may include the number of times it is selected as a candidate point of interest, which may be the number of times the point of interest (in response to a user search) is presented. The number of times presented may be the number of times presented in response to searches or navigation by all users.
The tag may in turn include the number of times the candidate point of interest was clicked for viewing by the user. The candidate points of interest are not limited to the current user but are clicked for viewing by all users.
The tag may also include the time when the candidate point of interest was accessed. The time may be aggregated into several hours of morning, noon, afternoon, evening, midnight, etc.
In addition, the tag may also include the proportion of access of the candidate point of interest by the local user and the foreign user. The judging basis of the local user and the foreign user can be determined according to the historical behavior analysis of the user. For example, in the past year, if more than a predetermined number of destinations in the search record and navigation record of the first user are in beijing, the first user may be determined to be a beijing user. In the past year, if more than a predetermined number of destinations in the search record and navigation record of the second user are in the open sea, the second user may be determined to be an open sea user. In the case that the Shanghai user accesses a certain interest point of Beijing, the interest point can be recorded as one access of the foreign user. The ratio of the number of accesses of the candidate interest point by the local user to the number of accesses of the foreign user can be used as the access ratio of the local user to the foreign user.
Different tags may be assigned different weights. The allocation mode can be predetermined, and can be determined by calculation by using a weight calculation model.
And inputting the labels of each candidate interest point and the weights of the labels into a pre-trained ranking model, so that the ranking of each candidate interest point can be obtained. The output of the ranking model may be the probability that each candidate point of interest is clicked, the probability that each candidate point of interest is presented, etc. And according to the result output by the sorting model, sorting the candidate interest points.
Through the scheme, after the candidate interest points are determined, the candidate interest points can be ranked according to the labels of the candidate interest points. So that the sorting result is more in line with the preference of the current user.
In one embodiment, the labeling of candidate points of interest includes: as at least one of the number of candidate points of interest, the number of times accessed, the time accessed, and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user.
Through the scheme, the accuracy of sequencing is improved by using the multi-dimensional labels of the candidate interest points.
As shown in connection with fig. 3, in one embodiment, the determination of a plurality of candidate points of interest from the historical points of interest involved in step S102 may include the sub-steps of:
s301: acquiring characteristics of historical interest points;
s302: and taking the points of interest with similar characteristics to the historical points of interest as candidate points of interest according to the similarity of the characteristics.
The historical behavior may include search records, navigation records, etc. of the current user. Through big data analysis, the historical behavior of the user on the map to the consumer interest points is sparse, and only interests are usually expressed for a period of time to a few consumer interest points, so that the interests of the current user need to be expanded and mined. The consumer interest point may be a restaurant, hotel, movie theater, park, or the like.
For example, if the search record or navigation record of the current user contains a certain hotpot shop, the hotpot shop can be used as a historical interest point. The hot pot store can be characterized by catering, hot pot, geographic location, average consumption amount, main dishes, etc.
For another example, if a park is included in the search record or navigation record of the current user, the park may serve as a historical point of interest. The park may be characterized as a point of interest, a red leaf theme park, a geographic location, etc.
According to the feature similarity, points of interest having the same or similar features as the historical points of interest may be considered candidate points of interest. The feature similarity may be points of interest of the same class, for example, a hotpot, and the class of hotpot may be a restaurant or a hot pot. Points of interest having the same vertical class as the hotpot store can be considered candidate points of interest.
As another example, a park may be characterized as a red leaf theme park, where points of interest of similar features may also be selected without points of interest of the same features. For example, a tulip theme park, ginkgo theme park, etc. may be selected.
In the above, taking the same or similar characteristics as an example, in the process of virtually determining the candidate interest points, multiple characteristics of the reference interest points can be integrated, so as to realize the best matching. Feature similarity may be matched according to Euclidean distance between features.
According to the scheme, under the condition that the historical interest points are few, the interest points can be expanded by utilizing the feature similarity, so that candidate interest points are obtained.
In one embodiment, the characteristics of the historical points of interest include: at least one of verticals, geographic location, average consumption price, and user rating.
Through the scheme, the number of candidate interest points can be increased by utilizing the multidimensional feature.
As shown in connection with fig. 4, in one embodiment, the determination of a plurality of candidate points of interest from the historical points of interest involved in step S102 may include the sub-steps of:
s401: acquiring historical operation behaviors of other users who access the historical interest points;
s402: and from the historical operation behaviors of other users, determining other points of interest which are accessed by other users after the historical points of interest are accessed, and determining the other points of interest as candidate points of interest.
After determining the historical interest point accessed by the current user, the database can be queried for the historical operation behaviors of other users who access the historical interest point.
For example, if the first user's historical point of interest is a hotpot, the database may be queried for other users who have also accessed the hotpot. The interest points visited by other users after visiting the hot pot store can be obtained, and the interest points visited by other users after visiting the hot pot store are determined to be candidate interest points.
It will be appreciated that other points of interest may also be accessed by other users prior to accessing the historical points of interest.
Through the scheme, the interest points can be expanded by utilizing the historical behaviors of the current user and other users. Namely, by searching for the historical behaviors of users with the same interest, the expansion of the interest points is realized.
Referring to fig. 5, the present application provides a training method for a point of interest recommendation model, which may include the following steps:
s501: for a plurality of interest point samples, obtaining labels and sequencing truth values of each interest point sample;
s502: determining the weight of the label of the interest point sample;
s503: obtaining a sequencing predicted value of each interest point sample according to the label of each interest point sample and the weight of the label by the interest point recommendation model to be trained;
s504: training the interest point recommendation model to be trained according to the sequencing predicted value and the sequencing true value until the errors of the sequencing predicted value and the sequencing true value are within the allowable range.
The point of interest sample may be a pre-labeled sample. For example, for a hot pot store, tags such as the number of times the hot pot store is a candidate point of interest, the number of times it is accessed, the time it is accessed, the proportion of accesses by local and foreign users, etc. may be counted.
Second, weights may be assigned for different tags. The weight distribution mode can be distributed by using a weight distribution calculation model, and can also be distributed by other modes.
In addition, the marked samples also comprise sorting true values. The ranking truth value may be the click rate, number of searches, etc. data.
And inputting the labels loaded with different weights into the interest point recommendation model to be trained, and obtaining the sequencing predicted value of each interest point sample. The ranking prediction value is compared to the ranking true value with an error. The error is counter-propagated in each layer of the point-of-interest recommendation model to be trained, and parameters of each layer are adjusted according to the error until the output of the point-of-interest recommendation model to be trained converges or reaches the expected effect. For example, it may be that the error of the ranking predicted value and the ranking true value is within an allowable range.
In one embodiment, the tag of the point of interest sample includes: at least one of the number of times presented, the number of times accessed, the time accessed, the proportion accessed by the local user and the foreign user.
Referring to fig. 6, the present application provides a recommendation device for an interest point, where the device may include:
a historical interest point determining module 601, configured to determine a historical interest point visited by the current user according to a historical behavior of the current user;
a candidate interest point determining module 602, configured to determine a plurality of candidate interest points according to the historical interest points;
a ranking module 603, configured to rank the plurality of candidate points of interest by using the tag of each candidate point of interest;
a recommending module 604, configured to recommend a plurality of candidate points of interest according to the ranking.
In one embodiment, the ordering module 603 may further include:
the label extraction sub-module is used for extracting a plurality of labels of each candidate interest point;
the weight determining sub-module is used for determining the weight of each tag of each candidate interest point;
and the sorting execution sub-module is used for inputting each label of each candidate interest point and the weight of each label into a pre-trained sorting model to obtain a sorting result of the candidate interest point.
In one embodiment, the labeling of candidate points of interest includes: as at least one of the number of candidate points of interest, the number of times accessed, the time accessed, and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user.
In one embodiment, the candidate point of interest determination module 602 may further include:
the characteristic acquisition sub-module is used for acquiring the characteristics of the historical interest points;
and the candidate interest point determination execution sub-module is used for taking the interest points with similar characteristics with the historical interest points as candidate interest points according to the similarity of the characteristics.
In one embodiment, the characteristics of the historical points of interest include: at least one of verticals, geographic location, average consumption price, and user rating.
In one embodiment, the candidate point of interest determination module 602 may further include:
the historical operation behavior acquisition module of other users is used for acquiring the historical operation behaviors of other users who access the historical interest points;
and the candidate interest point determination execution sub-module is used for determining other interest points which are accessed by other users after the other users access the historical interest points from the historical operation behaviors of the other users, and determining the other interest points as candidate interest points.
As shown in fig. 7, the present application provides a training device for a point of interest recommendation model, which may include:
the interest point sample information obtaining module 701 is configured to obtain, for a plurality of interest point samples, a label and a ranking truth value of each interest point sample;
a weight determining module 702 for determining the weight of the tag of the point of interest sample;
the ranking prediction value determining module 703 is configured to enable the point of interest recommendation model to be trained to obtain a ranking prediction value of each point of interest sample according to the tag of each point of interest sample and the weight of the tag;
and the training module 704 is configured to train the interest point recommendation model to be trained according to the ranking predicted value and the ranking true value until the error of the ranking predicted value and the ranking true value is within the allowable range.
In one embodiment, the tag of the point of interest sample includes: at least one of the number of times presented, the number of times accessed, the time accessed and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user.
FIG. 8 is a scene graph in which a recommendation method for points of interest may be implemented.
The interest capturing module is used for storing the behaviors of the user in the whole map, such as searching, navigation and the like, into the behavior log library and is used for learning the interest of the user in the interest points. Such as the user retrieving a jockey, navigating to a home, etc., will be recorded as a sequence of points of interest at this stage, with different actions (retrieving, navigating) being given different weights.
Illustrating: assuming that the user has searched for a plain, palace, or round for a period of time, the sequence of interest is recorded as:
[ Yihe garden, fet 1] [ Imperial palace, fet 2] [ Yuanqingyuan, fet 3]
Wherein the feat contains user operating characteristics for the point of interest, such as operating time, type of operation (search, navigation, etc.), user status (whether on the way or not), etc.
An interest expansion mining module: the behavior of users on maps for consumer interest points is usually sparse, and only interest is usually expressed for a few interest points in a period of time, so that the interests of the users need to be expanded and mined. And (5) performing interest expansion data mining through behavior association mining, a tag library and a content library, and performing offline storage. The online part can achieve the purpose of interest expansion only by carrying out query operation according to the interest point sequence of the user.
Behavior association mining proceeds mainly from two aspects: 1. and (5) association of the content. Recommending recommendation items similar to the interesting point of interest content for the user. When the user clicks the hot pot shop, the hot pot shop with the same price and a relatively close distance can be used as a recommendation. 2. Correlation of click behavior. It can be understood that most users search for which places after searching for the hot pot store, and take the places searched later as similar points of interest. And mining the corresponding behavior association recall according to the behavior association.
The content library may be used for ontology recall. The user searches for the hot pot store A, and the hot pot store A can be used as a recommendation, namely, the body recall. The tag library stores tags of each interest point, so that the similarity of the interest points can be determined according to the distance between the tags, namely, the content association recall is facilitated.
The ontology recall, the content association recall and the behavior association recall form the candidate interest points.
The interest ordering module is used for ordering the candidate interest points. And scoring the candidate interest points by using the LTR model, and finally outputting the sequencing result. The learning of the LTR model can be performed according to the content stored in the behavior log and the scene log. The scene log is used for storing scenes when the user acts, such as action time, place, driving or not, and the like. For learning of the LTR model, iteration may be performed by means of user behavior acquired in real time from big data. Thereby enabling the precision of the online sequencing model to meet the requirement.
According to embodiments of the present application, the present application also provides an electronic device, a readable storage medium and a computer program product.
FIG. 9 shows a schematic block diagram of an example electronic device 900 that may be used to implement an embodiment of the application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the applications described and/or claimed herein.
As shown in fig. 9, the apparatus 900 includes a computing unit 910 that may perform various suitable actions and processes according to a computer program stored in a Read Only Memory (ROM) 920 or a computer program loaded from a storage unit 980 into a Random Access Memory (RAM) 930. In the RAM 930, various programs and data required for the operation of the device 900 may also be stored. The computing unit 910, ROM 920, and RAM 930 are connected to each other by a bus 940. An input output (I/O) interface 950 is also connected to bus 940.
Various components in device 900 are connected to I/O interface 950, including: an input unit 960, such as a keyboard, mouse, etc.; an output unit 970 such as various types of displays, speakers, and the like; a storage unit 980, such as a magnetic disk, optical disk, etc.; and a communication unit 990 such as a network card, modem, wireless communication transceiver, etc. Communication unit 990 allows device 900 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks.
The computing unit 910 may be a variety of general purpose and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 910 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 910 performs the various methods and processes described above, such as a recommendation method for points of interest and/or a training method for a recommendation model for points of interest. For example, in some embodiments, the point of interest recommendation method and/or the training method of the point of interest recommendation model may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 980. In some embodiments, some or all of the computer program may be loaded and/or installed onto device 900 via ROM 920 and/or communication unit 990. When the computer program is loaded into the RAM 930 and executed by the computing unit 910, one or more steps of the point-of-interest recommendation method and/or the training method of the point-of-interest recommendation model described above may be performed. Alternatively, in other embodiments, the computing unit 910 may be configured to perform the point of interest recommendation method and/or the training method of the point of interest recommendation model in any other suitable manner (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present application may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution disclosed in the present application can be achieved, and are not limited herein.
The above embodiments do not limit the scope of the present application. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the scope of the present application.
Claims (14)
1. A training method of a point of interest recommendation model comprises the following steps:
for a plurality of interest point samples, acquiring a label and a ranking truth value of each interest point sample, wherein the label of the interest point sample comprises: at least one of the number of times presented, the number of times accessed, the time accessed and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user, wherein the ranking truth value comprises at least one of click rate and search frequency;
determining the weight of the label of the interest point sample;
obtaining a sequencing predicted value of each interest point sample according to the label of each interest point sample and the weight of the label by an interest point recommendation model to be trained;
and training the interest point recommendation model to be trained according to the sequencing predicted value and the sequencing true value until the error of the sequencing predicted value and the sequencing true value is within an allowable range.
2. A method of recommending points of interest, comprising:
determining historical interest points visited by the current user according to the historical behaviors of the current user;
determining a plurality of candidate interest points according to the historical interest points;
sorting the plurality of candidate interest points by using the label of each candidate interest point;
recommending the candidate interest points according to the ranking;
the ranking the plurality of candidate interest points by using the label of each candidate interest point includes:
extracting a plurality of labels of each candidate interest point;
determining the weight of each tag of each candidate interest point;
and for each candidate interest point, inputting each label of the candidate interest point and the weight of each label into the interest point recommendation model obtained by training the interest point recommendation model by adopting the training method of the interest point recommendation model as claimed in claim 1, and obtaining the sequencing result of the candidate interest points.
3. The method of claim 2, wherein the labeling of the candidate points of interest comprises: as at least one of the number of candidate points of interest, the number of times accessed, the time accessed, and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user.
4. A method according to any one of claims 2 to 3, wherein said determining historical points of interest visited by the current user based on the current user's historical behaviour comprises:
acquiring characteristics of the historical interest points;
and taking the points of interest with similar characteristics to the historical points of interest as candidate points of interest according to the similarity of the characteristics.
5. The method of claim 4, the characteristics of the historical points of interest comprising: at least one of verticals, geographic location, average consumption price, and user rating.
6. A method according to any one of claims 2 to 3, wherein determining a plurality of candidate points of interest from the historical points of interest comprises:
acquiring historical operation behaviors of other users accessing the historical interest points;
and determining other interest points which are accessed by the other users after the other users access the historical interest points from the historical operation behaviors of the other users, and determining the other interest points as candidate interest points.
7. A training device for a point of interest recommendation model, comprising:
the interest point sample information acquisition module is used for acquiring labels and sequencing truth values of each interest point sample for a plurality of interest point samples, wherein the labels of the interest point samples comprise: at least one of the number of times presented, the number of times accessed, the time accessed and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user, wherein the ranking truth value comprises at least one of click rate and search frequency;
the weight determining module is used for determining the weight of the label of the interest point sample;
the sequencing predicted value determining module is used for enabling the interest point recommendation model to be trained to obtain the sequencing predicted value of each interest point sample according to the label of each interest point sample and the weight of the label;
and the training module is used for training the interest point recommendation model to be trained according to the sequencing predicted value and the sequencing true value until the error of the sequencing predicted value and the sequencing true value is within an allowable range.
8. A point of interest recommendation device, comprising:
the historical interest point determining module is used for determining historical interest points visited by the current user according to the historical behaviors of the current user;
the candidate interest point determining module is used for determining a plurality of candidate interest points according to the historical interest points;
the sorting module is used for sorting the plurality of candidate interest points by using the label of each candidate interest point;
a recommending module, configured to recommend the plurality of candidate interest points according to the ranking;
wherein, the sequencing module includes:
the label extraction sub-module is used for extracting a plurality of labels of each candidate interest point;
a weight determining sub-module, configured to determine a weight of each tag of each candidate interest point;
and the ranking execution sub-module is used for inputting each label of the candidate interest points and the weight of each label into the interest point recommendation model obtained by training the interest point recommendation model according to the training method of the interest point recommendation model in claim 1 to obtain the ranking result of the candidate interest points.
9. The apparatus of claim 8, wherein the tag of the candidate point of interest comprises: as at least one of the number of candidate points of interest, the number of times accessed, the time accessed, and the ratio of the number of times accessed by the local user to the number of times accessed by the foreign user.
10. The apparatus of any of claims 8 to 9, wherein the candidate point of interest determination module comprises:
the characteristic acquisition sub-module is used for acquiring the characteristics of the historical interest points;
and the candidate interest point determination execution sub-module is used for taking the interest points with similar characteristics with the historical interest points as candidate interest points according to the similarity of the characteristics.
11. The apparatus of claim 8, the characteristics of the historical points of interest comprising: at least one of verticals, geographic location, average consumption price, and user rating.
12. The apparatus of any of claims 8 to 9, wherein the candidate point of interest determination module comprises:
the historical operation behavior acquisition module is used for acquiring the historical operation behaviors of other users who access the historical interest points;
and the candidate interest point determination execution sub-module is used for determining other interest points which are accessed by the other users after the other users access the historical interest points from the historical operation behaviors of the other users, and determining the other interest points as candidate interest points.
13. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 6.
14. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011554561.8A CN112612957B (en) | 2020-12-24 | 2020-12-24 | Recommendation method of interest points and training method and device of recommendation model of interest points |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011554561.8A CN112612957B (en) | 2020-12-24 | 2020-12-24 | Recommendation method of interest points and training method and device of recommendation model of interest points |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112612957A CN112612957A (en) | 2021-04-06 |
| CN112612957B true CN112612957B (en) | 2023-09-29 |
Family
ID=75245208
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011554561.8A Active CN112612957B (en) | 2020-12-24 | 2020-12-24 | Recommendation method of interest points and training method and device of recommendation model of interest points |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112612957B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113139139B (en) * | 2021-04-28 | 2023-09-22 | 北京百度网讯科技有限公司 | Method, device, electronic equipment and medium for determining boarding point |
| CN113536152B (en) | 2021-06-30 | 2024-02-09 | 北京百度网讯科技有限公司 | Map interest point display method and device and electronic equipment |
| CN113536120B (en) * | 2021-07-14 | 2023-08-01 | 北京百度网讯科技有限公司 | Point-of-interest recall method and device based on user behavior |
| CN113407849B (en) * | 2021-07-15 | 2024-06-14 | 百果园技术(新加坡)有限公司 | Coarse ordering method and device |
| CN113569149B (en) * | 2021-07-30 | 2024-01-19 | 拉扎斯网络科技（上海）有限公司 | Information processing method and device and electronic equipment |
| CN113868532B (en) * | 2021-09-30 | 2022-11-01 | 北京百度网讯科技有限公司 | Location recommendation method and device, electronic equipment and storage medium |
| CN113822594B (en) * | 2021-09-30 | 2024-08-27 | 阿里巴巴创新公司 | Point-of-interest score determination method, electronic device and computer program product |
| CN114708008B (en) * | 2021-12-30 | 2024-12-03 | 北京有竹居网络技术有限公司 | A promotional content processing method, device, equipment, medium and product |
| CN114443959B (en) * | 2022-01-26 | 2025-04-18 | 北京百度网讯科技有限公司 | Method, device, electronic device and readable storage medium for recommending points of interest |
| CN114662020A (en) * | 2022-04-29 | 2022-06-24 | 深圳依时货拉拉科技有限公司 | Method and device for point of interest recommendation based on contour of point of interest |
| CN115344175A (en) * | 2022-08-17 | 2022-11-15 | 江苏满运物流信息有限公司 | Map navigation method, apparatus, device and medium |
| CN115470401B (en) * | 2022-08-22 | 2025-07-15 | 北京百度网讯科技有限公司 | Resource recommendation method, device, electronic device and storage medium |
| CN115858954B (en) * | 2022-12-29 | 2024-09-17 | 腾讯科技（深圳）有限公司 | Method and device for recommending interest points, electronic equipment and storage medium |
| CN116757348A (en) * | 2023-07-05 | 2023-09-15 | 黑龙江省辰源投资管理有限公司 | Travel information intelligent planning management system and method based on artificial intelligence |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105227971A (en) * | 2015-07-09 | 2016-01-06 | 传成文化传媒（上海）有限公司 | A method and system for recommending information on TVs used in hotels |
| CN108804619A (en) * | 2018-05-31 | 2018-11-13 | 腾讯科技（深圳）有限公司 | Interest preference prediction technique, device, computer equipment and storage medium |
| WO2019187421A1 (en) * | 2018-03-29 | 2019-10-03 | ソニー株式会社 | Information processing device, information processing method, and program |
| CN110941739A (en) * | 2018-09-22 | 2020-03-31 | 北京微播视界科技有限公司 | Media file recommendation method and device, media file server and storage medium |
| CN111008321A (en) * | 2019-11-18 | 2020-04-14 | 广东技术师范大学 | Recommendation method and device based on logistic regression, computing equipment and readable storage medium |
| CN111831899A (en) * | 2019-09-24 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Navigation interest point recommendation method and device and server |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9824469B2 (en) * | 2012-09-11 | 2017-11-21 | International Business Machines Corporation | Determining alternative visualizations for data based on an initial data visualization |
-
2020
- 2020-12-24 CN CN202011554561.8A patent/CN112612957B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105227971A (en) * | 2015-07-09 | 2016-01-06 | 传成文化传媒（上海）有限公司 | A method and system for recommending information on TVs used in hotels |
| WO2019187421A1 (en) * | 2018-03-29 | 2019-10-03 | ソニー株式会社 | Information processing device, information processing method, and program |
| CN108804619A (en) * | 2018-05-31 | 2018-11-13 | 腾讯科技（深圳）有限公司 | Interest preference prediction technique, device, computer equipment and storage medium |
| CN110941739A (en) * | 2018-09-22 | 2020-03-31 | 北京微播视界科技有限公司 | Media file recommendation method and device, media file server and storage medium |
| CN111831899A (en) * | 2019-09-24 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Navigation interest point recommendation method and device and server |
| CN111008321A (en) * | 2019-11-18 | 2020-04-14 | 广东技术师范大学 | Recommendation method and device based on logistic regression, computing equipment and readable storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于多类型情景信息的兴趣点推荐模型;胡德敏;杨晨;;计算机应用研究(第06期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112612957A (en) | 2021-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112612957B (en) | Recommendation method of interest points and training method and device of recommendation model of interest points | |
| CN111538904B (en) | Method and device for recommending interest points | |
| CN112100524B (en) | Information recommendation method, device, equipment and storage medium | |
| CN112559879B (en) | Interest model training method, interest point recommending method, device and equipment | |
| CN113656698B (en) | Training method and device for interest feature extraction model and electronic equipment | |
| CN105208113A (en) | Information pushing method and device | |
| CN111666461B (en) | Method, apparatus, device and computer storage medium for retrieving geographic location | |
| WO2022134478A1 (en) | Route recommendation method and apparatus, electronic device, and storage medium | |
| CN111666292B (en) | Similarity model establishment method and device for retrieving geographic position | |
| CN112989146B (en) | Method, apparatus, device, medium and program product for recommending resources to target user | |
| US20120084226A1 (en) | Measuring or estimating user credibility | |
| CN112528145B (en) | Information recommendation method, device, equipment and readable storage medium | |
| CN113536152B (en) | Map interest point display method and device and electronic equipment | |
| CN109241403A (en) | Item recommendation method, device, machinery equipment and computer readable storage medium | |
| CN105447005B (en) | Object pushing method and device | |
| CN112541123A (en) | Map region recommendation method, device, equipment and storage medium | |
| CN111339423B (en) | User-based travel city pushing method, system, equipment and storage medium | |
| CN109977292A (en) | Searching method, calculates equipment and computer readable storage medium at device | |
| CN113792212A (en) | Multimedia resource recommendation method, device, equipment and storage medium | |
| CN112559901B (en) | Resource recommendation method and device, electronic equipment, storage medium and computer program product | |
| US20120271844A1 (en) | Providng relevant information for a term in a user message | |
| US11976935B2 (en) | Route recommendation method, electronic device, and storage medium | |
| CN113326450B (en) | Point-of-interest recall method and device, electronic equipment and storage medium | |
| CN112052402B (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN113901056A (en) | Interface recommendation method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |