WO2018227773A1 - 地点推荐方法、装置、计算机设备和存储介质 - Google Patents

地点推荐方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2018227773A1
WO2018227773A1 PCT/CN2017/099735 CN2017099735W WO2018227773A1 WO 2018227773 A1 WO2018227773 A1 WO 2018227773A1 CN 2017099735 W CN2017099735 W CN 2017099735W WO 2018227773 A1 WO2018227773 A1 WO 2018227773A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
check
location
recommended
query
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.)
Ceased
Application number
PCT/CN2017/099735
Other languages
English (en)
French (fr)
Inventor
王健宗
黄章成
吴天博
肖京
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Publication of WO2018227773A1 publication Critical patent/WO2018227773A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/40—Business processes related to social networking or social networking services
    • G06Q10/42—Determination of affinities or common interests between users
    • 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

Definitions

  • personalized recommendations can help users filter out information that users are not interested in from rich and cumbersome data, and better discover user preferences to increase user activity in social networks.
  • the traditional personalized location recommendation method mostly analyzes the user's historical trajectory data, obtains the user's location preference, and then recommends the location similar to the preference to the user.
  • the location recommendation method based on the user's own trajectory data has the following drawbacks: First, since the amount of data of the user's own trajectory is small and single, the recommended location is relatively simple. Secondly, there may be places in the user history track data that the user does not like. Therefore, personalized recommendation based on the user's own historical track data cannot ensure accurate user preference.
  • a location recommendation method is provided.
  • a location recommendation method comprising:
  • a location recommending device comprising:
  • a request receiving module configured to receive a location recommendation request sent by the user terminal, where the location recommendation request carries the query user identifier
  • the check-in data search module is configured to search for the check-in data of the query user corresponding to the query user identifier, where the location social network platform generates the check-in data set according to the historical check-in behavior of the user, and the check-in data of each user in the check-in data set All include check-in locations;
  • An associated user determining module configured to search for an associated user of the querying user in the check-in data set, where the associated user has at least one check-in place and a check-in place of the querying user;
  • a similar user set determining module configured to calculate a similarity between the query user and each of the associated users, and determine a similar user set corresponding to the query user according to the calculated similarity
  • a check-in place collection determining module configured to determine a check-in place set corresponding to the similar user set, where the check-in place set includes a check-in place signed by all related users in the similar user set;
  • a recommended location determining module configured to remove the check-in location in the set of check-in locations that coincides with the querying user, obtain a set of recommended locations, and push the recommended location included in the set of recommended locations to the querying user terminal.
  • a computer apparatus comprising a memory and a processor, the memory storing computer executable instructions, the instructions being executed by the processor, causing the processor to perform the following steps:
  • the check-in data includes the check-in location
  • One or more non-volatile readable storage media storing computer-executable instructions, the instructions being executed by one or more processors, such that the one or more processors perform the following steps:
  • 1 is an application environment diagram of a location recommendation method in an embodiment
  • FIG. 2 is a flow chart of a method for recommending a place in an embodiment
  • FIG. 3 is a flowchart of calculating a similarity between a query user and an associated user in an embodiment
  • Figure 4 is a flow chart involved in the location recommendation step in one embodiment
  • Figure 5 is a block diagram showing the structure of a location recommendation device in an embodiment
  • FIG. 6 is a schematic diagram showing the internal structure of a server in an embodiment.
  • an application environment diagram of a location recommendation method including an inquiry terminal 110 and a server 120.
  • the query terminal 110 can communicate with the server 120 over a network.
  • the query terminal 110 may be at least one of a smartphone, a tablet, a notebook, and a desktop computer, but is not limited thereto.
  • the server 120 is a server or a server cluster of a Location-based Social Network (LBSN), and the user's check-in data is stored in the database.
  • LBSN Location-based Social Network
  • the server 120 receives the location recommendation request sent by the user terminal, searches for the check-in data of the query user from the check-in database according to the carried query user identifier, and then searches for the associated user that has the overlapping check-in location with the query user from the check-in database.
  • the server 120 calculates the similarity between the query user and the associated user according to the check-in data of the query user and the check-in data of the associated user, and determines the associated user whose similarity meets the set condition as the member of the similar user set corresponding to the query user.
  • the members of the similar user set have similar location preferences as the querying user, and the check-in place of the similar user centralized member can better fit the querying user's preference, and the check-in location is relative to the querying user's history. Points have a certain diversity.
  • FIG. 2 is a schematic flow chart of a location recommendation method according to an embodiment of the present invention. It should be understood that although the various steps in the flowchart of FIG. 2 are sequentially displayed as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Except as explicitly stated herein, the execution of these steps is not strictly limited, and may be performed in other sequences. Moreover, at least some of the steps in FIG. 2 may include a plurality of sub-steps or stages, which are not necessarily performed sequentially, but may be alternated or alternated with at least a part of other steps or sub-steps or stages of other steps. carried out.
  • a location recommendation method is provided.
  • the method is illustrated in the server 120 in FIG. 1, and specifically includes the following steps:
  • Step S202 Receive a location recommendation request sent by the user terminal, where the location recommendation request carries the query user identifier.
  • the terminal device that can communicate with the location social network platform server is installed in the query user terminal, and the user terminal is queried to send a location recommendation request to the server through the terminal application.
  • the recommendation request can be sent by clicking on the "Location Recommendations" button in the application interface.
  • the user After the terminal logs in to the server, the user sends a recommendation request to the server by using a regular shaking terminal body.
  • the server receives the user login platform request, it is deemed to query the terminal where the user is located to send a location recommendation request to the server, that is, the server performs location recommendation for each logged-in user.
  • the above location social networking platform can be Foursquare, Gowalla or Facebook Places.
  • the user's check-in time, check-in location, and evaluation content made to the location are included.
  • the evaluation content made by the user on the location may include an evaluation in a text form, and may also include a rating evaluation.
  • Step S204 Search for the check-in data of the query user corresponding to the query user identifier, where the location social network platform generates the check-in data set according to the historical check-in behavior of the user, and the check-in data of each user in the check-in data set includes the check-in place.
  • the server searches for the corresponding check-in data in the check-in data set according to the user ID of the querying user.
  • the check-in data set is a set of check-in data generated according to the check-in behavior of the user in the platform, and the user's check-in behavior is a check-in request sent to the server at a certain place during the historical time period.
  • the check-in request carries the user's evaluation information about the check-in location.
  • the server generates the check-in data according to the check-in behavior of the user, including: obtaining the location information of the user when the check-in is located, and locating the check-in place according to the location information, such as a restaurant, a tourist scenic spot, and the like.
  • the check-in data further includes generating numerical rating information for the check-in location based on the evaluation information entered by the user.
  • step S206 the associated user of the querying user is searched in the check-in data set, and the associated user has at least one check-in place coincident with the check-in place of the querying user.
  • the server finds whether there is a check-in data that coincides with the check-in location of the query user in the check-in data set. If yes, the user corresponding to the checked-in data is defined as the associated user of the query user.
  • the associated user and the querying user may have a check-in place coincident, or multiple check-in locations may coincide.
  • the check-in data of the query user u is: a, b, c
  • the check-in data of the user v is: a, d, e
  • the check-in data of the user w is: b, a, f,.
  • the user v has a location a that has been visited by the query user u. Therefore, the user v is the associated user who queries the user u.
  • the user w and the query user u have two locations that have been visited together, respectively b and a. Therefore, the user w is an associated user who queries the user u.
  • Step S208 Calculate the similarity between the query user and each associated user, and determine a similar user set corresponding to the query user according to the calculated similarity.
  • the associated user of the querying user determined according to step S206 may be a large number of user groups.
  • this step filters the determined associated users, and determines and selects the user's location preferences from the associated users.
  • a similar group of users that is, a set of similar users that determine the querying user.
  • the specific method for determining the similar user set of the querying user is: calculating the similarity between the two according to the check-in data of the querying user and the associated user, and selecting the sorting according to the order of similarity in the order of similarity.
  • the associated user of the previous set number of similarities is the similar user set of the querying user. That is, the associated users corresponding to the first N largest similarities are selected as the similar user set of the querying user.
  • Step S210 Determine a set of check-in locations corresponding to the set of similar users, where the set of check-in locations includes A similar user collects the check-in locations that all associated users have checked in.
  • Step S212 Remove the check-in place that coincides with the query user in the set of check-in places, obtain a set of recommended places, and push the recommended place included in the set of recommended places to the query user terminal.
  • the check-in location set corresponding to the similar user set is determined according to the check-in data of all associated users included in the similar user set. That is, the check-in locations visited by all associated users in a similar user set can be found in the corresponding check-in place set.
  • the associated users in a similar user set are: associated users A: a, d, e; associated users B: b, a, f; and associated users C: c, b, e, then the similar user set
  • the corresponding check-in place set is ⁇ a, b, c, d, e, f ⁇ .
  • the check-in place where the user has checked in is removed from the check-in place set corresponding to the similar user set, and the recommended place set is obtained. If the check-in location set of the query user is ⁇ a, b, c), the recommended place set is ⁇ d, e, f ⁇ , and the recommended place in the recommended place set is pushed to the query user terminal.
  • the user in the similar user set of the query user has a similar location preference to the query user, and the location visited by the user in the similar user set can fit the query user's preference with a certain probability.
  • the sign-in location corresponding to the similar user set is used as the basis of the recommended location, and the diversity of the users determines that the recommended location is also more diverse.
  • the recommended location is not limited to the same type of location that queries the location visited by the user itself, but may be to some extent fit other types of locations that query the user's preferences.
  • the check-in data further includes generating numerical rating information for the check-in location based on the rating information entered by the user.
  • the check-in data of the querying user may include (a, 0.8), (b, 0.5), (c, 0.3) 3 check-in data.
  • the a, b, and c in the check-in data are the check-in locations of the query user, and the value in each data is the score of the query user for the check-in place.
  • Each check-in data of the query user corresponds to a specific check-in time. When the query user signs in to a check-in location at a different check-in time, multiple check-in data will be generated. If the user is checked in at time t 1 and t 2, the user is checked in at location a. Two check-in data will be generated, such as (a, 0.8), (a, 0.9).
  • step S208 calculating a similarity between the query user and each associated user, and determining a similar user set corresponding to the query user according to the calculated similarity, including the following steps:
  • Step S302 Calculate the similarity between each associated user and the querying user, wherein the querying user The more concentrated the check-in places with the associated users and the closer the scores to the common check-in places, the larger the calculated similarity value.
  • the determination of whether the check-in location of the user and the associated user is centralized depends on the following factors: the sum of the times of signing in to the same place (considering the case where one of the users signs in to the common check-in place multiple times) and the sum of the times of signing in the non-shared place. The greater the sum of the times of signing in to the same place and the smaller the total number of times of signing in the non-co-signing place, the more concentrated the check-in place of the associated user and the querying user.
  • the set of the check-in location of the query user is ⁇ a, b, c, a, d, e, f, e, h ⁇
  • the set of the check-in locations of the associated user A is ⁇ a, m, i, l, m, b, f, k, h ⁇
  • the collection of the location of the associated user B is ⁇ a, b, e, o, f, k, m ⁇
  • the total number of times that the associated user A and the querying user check in to the same place is 5, which are a, a, b, f, and h, and the number of times of signing in the non-shared place is 9, c, d, e, e, m, i, respectively. l, m, k; the total number of times that the associated user B and the querying user check in to the same place are 6, respectively, a, b, a, e, e, f, and the number of times of signing in the non-shared place is 6, respectively, c, d, h, o, m, k. It can be seen from the above analysis that the location of the check-in between the associated user B and the querying user is more concentrated.
  • whether the scores of the common check-in places are relatively close can be determined by calculating the variance value or the standard deviation or the absolute value of the score difference of the scores of the querying user and the associated user for the same check-in place.
  • the absolute value of the score difference as an example, for example, the locations of the query user and the associated user are a and b respectively, and the scores of the two are ⁇ 0.6, 0.8 ⁇ , ⁇ 0.5, 0.9 ⁇ , respectively, and the scores of the two are respectively
  • the closeness is
  • Step S304 The related users whose similarity is greater than the set threshold are grouped into a similar user set of the querying user.
  • the similarity between the query user and the associated user is calculated, and the associated user whose similarity is greater than the set threshold is regarded as a member of the similar user set of the query user.
  • the size of the threshold may be preset, for example, may be 0.8. Associated users with a similarity greater than 0.8 constitute a similar set of users for the querying user. If the size of the similar user set member determined according to the preset threshold is small (for example, if the determined number of related users in the similar user set is less than 2), the size of the set threshold is adjusted to re-determine the similar user set.
  • the server presets a plurality of levels of similarity thresholds, such as an accurate threshold (eg, 0.8), standard threshold (eg 0.6), extensive threshold (eg 0.4), precise threshold > standard threshold > extensive threshold.
  • an accurate threshold eg, 0.8
  • standard threshold eg 0.6
  • extensive threshold eg 0.4
  • the calculated similarity can more reflect the degree of similarity between the associated user and the querying user, so that the recommendation of the location according to the similar user can be more appropriately posted. User's preference for location.
  • the similarity between each associated user and the querying user can be calculated by the following formula:
  • u and v represent the query user and the associated user respectively;
  • sim(u, v) is the similarity between the associated user and the querying user;
  • R ui and R vi are the scores of the query user and the associated user respectively for the location i;
  • r j is the number of times the query user or the associated user checks the location j;
  • r max is the location social
  • the number of times of sign-in corresponding to the sign-in location with the most sign-on by any user in the network platform is substantially a fixed value for the social network platform r max at the same location, and its role is to perform normalization of the score.
  • the similarity calculated by equation (1) not only measures the set of places that two users have visited together. Moreover, the location of other non-common visits, that is, the degree of dispersion (or concentration) of the two places of visit, and the consideration of the user's scoring factors on the place of check-in are considered, so that the similarity of the calculation can accurately evaluate the two. Whether the location preferences are similar.
  • step S210 in addition to the check-in location in the set of check-in locations that coincides with the querying user, obtaining a set of recommended locations, and pushing the recommended locations included in the set of recommended locations to the querying user terminal include:
  • Step S402 Remove the check-in place that coincides with the query user in the set of check-in places, and obtain a set of places to be recommended.
  • the location where all the query users have not checked in in the similar user set is the set of places to be recommended.
  • the place recommended to the querying user should be a place where the user has not been visited.
  • the location to be recommended by the querying user is extracted from the set of check-in locations corresponding to the similar user set.
  • the associated users in the similar user set have certain location preference similarities with the query users.
  • the query based on the check-in locations corresponding to the similar user sets is recommended to the query user preferences to a certain extent.
  • Step S404 Calculate the interest degree of each of the to-be-recommended locations in the set of the to-be-recommended locations, and push the to-be-recommended locations with the interest degree greater than the set threshold to the querying user terminal; wherein the interest degree is collected by querying users and similar users. The similarity between the associated users who sign in to the recommended location and the rating of the associated user to the recommended location are calculated.
  • the determined location to be recommended is further accurately selected to select the location that best fits the true preference of the querying user. Specifically: calculating the degree of interest between the location to be recommended and the querying user. The higher the interest between the recommended location and the querying user, the higher the fit between the location and the querying user preferences.
  • the related users who have checked in to the recommended location are first searched in a similar user set. Then, according to the similarity between the searched related user and the querying user calculated in step S208 and the score of the associated user on the check-in place, the degree of interest of the querying user and the to-be-queried place is calculated. That is, the relationship between the user and the location is obtained by the similarity relationship between the associated user and the querying user and the rating relationship between the associated user and the location.
  • the degree of interest between the location A to be recommended and the querying user is calculated.
  • the related users who have visited the to-be-recommended location A in a similar user set are respectively associated users u 1 , u 2 , and u 3 .
  • the respective check-in data of the associated users u 1 , u 2 , and u 3 includes their rating information for the recommended location A, and the query user is calculated according to the calculated similarity between the associated user and the querying user and the score of the associated user to the recommended location A.
  • the degree of interest of the recommended location A wherein the higher the rating of the associated user to the recommended location, the higher the similarity between the associated user and the querying user, the higher the interest of the recommended location and the querying user.
  • the degree of interest between the querying user and the place to be recommended may be calculated by the following formula (2):
  • u is the querying user
  • j is the determined to-be-recommended location in the set of to-be-recommended locations
  • U is the similar user set of the querying user
  • u k is the associated user of the similarly-intended centralized sign-to-recommended location j
  • sim(u , u k ) is the similarity between the query user u and the associated user u k , The rating of the recommended location j for the associated user u k .
  • the formula (2) can calculate the degree of interest of the querying user and the place to be recommended by using the relationship between the querying user and the associated user and the location to be recommended, and the degree of interest can well reflect the degree of interest of the querying user in the recommended location. Pushing the to-be-recommended location with a greater degree of interest to the querying user terminal, so that the pushed location is more suitable for the user's own real preference, that is, accurate pushing for the user is realized.
  • a location recommendation device comprising:
  • the request receiving module 502 is configured to receive a location recommendation request sent by the query user terminal, and recommend the location.
  • the request carries the query user ID.
  • the check-in data search module 504 searches for the check-in data of the query user corresponding to the query user identifier, wherein the location social network platform generates the check-in data set according to the historical check-in behavior of the user, and the check-in data of each user in the check-in data set includes the check-in place.
  • the associated user determining module 506 searches for the associated user of the querying user in the check-in data set, and the associated user has at least one check-in place coincident with the check-in location of the querying user.
  • the similar user set determining module 508 is configured to calculate a similarity between the query user and each associated user, and determine a similar user set corresponding to the query user according to the calculated similarity.
  • the check-in place collection determining module 510 is configured to determine a check-in place set corresponding to the similar user set, where the check-in place set includes the check-in place signed by all the associated users in the similar user set.
  • the recommended location determining module 512 is configured to remove the check-in place in the set of check-in locations that coincides with the querying user, obtain a set of recommended locations, and push the recommended locations included in the set of recommended locations to the querying user terminal.
  • the check-in data further includes a rating of the check-in location;
  • the similar user set determining module 508 is further configured to calculate a similarity between each associated user and the querying user, wherein the querying user and the associated user are checked in. The more concentrated the check-in place is, the closer the score to the common check-in place is, the larger the similarity value is calculated; the related users whose similarity is greater than the set threshold constitute a similar user set of the query user.
  • the similarity between each associated user and the querying user is calculated by the following formula:
  • u and v represent the query user and the associated user respectively;
  • sim(u, v) is the similarity between the associated user and the querying user;
  • R ui and R vi are the scores of the query user and the associated user respectively for the location i;
  • r max is the location of the check-in location of the location social network platform that is most frequently signed by any user. The number of check-ins.
  • the recommended location determining module 512 is further configured to remove the check-in place in the set of check-in locations that coincides with the querying user, obtain a set of locations to be recommended, and calculate an interest of each of the to-be-recommended locations in the set of the querying user and the to-be-recommended location.
  • the degree to be recommended is pushed to the querying user terminal, where the degree of interest is the degree of similarity between the user who is in the recommended location by the query user and the similar user, and the associated user treats the recommendation.
  • the rating of the location is calculated.
  • the formula for calculating the degree of interest of the user and the place to be recommended is:
  • u is the querying user
  • j is the determined to-be-recommended location in the set of to-be-recommended locations
  • U is the similar user set of the querying user
  • u k is the associated user of the similarly-intended centralized sign-to-recommended location j
  • sim(u , u k ) is the similarity between the query user u and the associated user u k , The rating of the recommended location j for the associated user u k .
  • the data distribution apparatus in the various embodiments described above may be implemented in the form of a computing program, and the computer executable instructions corresponding to the computer program may be executed on a computer device as shown in FIG.
  • the computer device can be a physical server or a server cluster composed of multiple servers.
  • Its internal structure includes: a processor connected through a system bus, a non-volatile storage medium, an internal memory, and a network interface.
  • the non-volatile storage medium of the computer device stores an operating system, a database, and at least one of the above-described computer-executable instructions implemented by the data distribution apparatus.
  • the database is used to store data, such as storing user's check-in data.
  • the processor is used to provide computing and control capabilities to support the operation of the entire computer device.
  • the internal memory provides an environment for the operation of an operating system in a non-volatile storage medium and computer-executable instructions for implementing data distribution.
  • the network interface is used for communication connection with the query terminal.
  • FIG. 6 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the server to which the solution of the present application is applied, and a specific server. More or fewer components than those shown in the figures may be included, or some components may be combined, or have different component arrangements.
  • the above network interface may be an Ethernet card or a wireless network card.
  • the above modules may also be embedded in hardware or independent of the computer device described above. It may also be stored in the memory of the differential distribution server in the form of software as described above, so that the processor calls to perform the operations corresponding to the above respective modules.
  • the processor can be a central processing unit (CPU), a microprocessor, a microcontroller, or the like.
  • one or more non-volatile readable storage media storing computer-executable instructions are provided, the instructions being executed by one or more processors, causing one or more processors to perform the All or part of the process in the embodiment method.
  • the computer executable instructions described above are computer executable instructions corresponding to a computer program implemented by all or part of the processes of the various embodiments described above.
  • the program can be stored in a computer readable storage medium, such as the present invention.
  • the program can be stored in a non-volatile readable storage medium of the computer system and executed by at least one processor in the computer system to implement a process comprising an embodiment of the methods described above.
  • the non-volatile readable storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种地点推荐方法，包括：接收查询用户终端发送的地点推荐请求，地点推荐请求中携带查询用户标识；查找与查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，签到数据集中每个用户的签到数据均包括签到地点；在签到数据集中查找查询用户的关联用户；计算查询用户与每个关联用户之间的相似度，并根据计算的相似度确定查询用户对应的相似用户集；确定相似用户集所对应的签到地点集合；去除签到地点集合中与查询用户重合的签到地点，得到推荐地点集合，将推荐地点集合中所包含的推荐地点推送至查询用户终端。

Description

地点推荐方法、装置、计算机设备和存储介质
本申请要求于2017年06月12日提交中国专利局、申请号为201710439338.0、发明名称为“地点推荐方法、装置、服务器和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
背景技术
在大数据的信息爆炸时代，个性化的推荐可以帮助用户从丰富而冗杂的数据中过滤掉用户不感兴趣的信息，更好的发现用户的偏好从而增加用户在社交网络中的活跃度。
传统的个性化位置推荐方法大都是根据用户历史轨迹数据进行分析，得到用户位置偏好，进而向用户推荐与偏好相类似的位置。这种基于用户自身轨迹数据进行的位置推荐方式存在如下缺陷：首先，由于用户自身轨迹数据量少且单一，推荐的位置也会比较单一。其次，用户历史轨迹数据中可能存在用户并不喜好的地点，因此，基于用户自身历史轨迹数据进行个性化推荐不能确保精准的贴合用户偏好。
发明内容
根据本申请公开的各种实施例，提供一种地点推荐方法、装置、计算机设备及存储介质。
一种地点推荐方法，所述方法包括：
接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有 一个签到地点与所述查询用户的签到地点重合；
计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；
去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
一种地点推荐装置，所述装置包括：
请求接收模块，用于接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
签到数据查找模块，用于查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
关联用户确定模块，用于在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
相似用户集确定模块，用于计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
签到地点集合确定模块，用于确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；
推荐地点确定模块，用于去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
一种计算机设备，包括存储器和处理器，所述存储器中存储有计算机可执行指令，所述指令被所述处理器执行时，使得所述处理器执行以下步骤：
接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户 的签到数据均包括签到地点；
在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；及
去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
一个或者多个存储有计算机可执行指令的非易失性可读存储介质，所述指令被一个或者多个处理器执行，使得所述一个或者多个处理器执行以下步骤：
接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；及
去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征、目的和优点将从说明书、附图以及权利要求书变得明显。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为一个实施例中地点推荐方法的应用环境图；
图2为一个实施例中地点推荐方法的流程图；
图3为一个实施例中计算查询用户与关联用户相似度所涉及的流程图；
图4为一个实施例中地点推荐步骤所涉及的流程图；
图5为一个实施例中地点推荐装置的结构框图；
图6为一个实施例中服务器的内部结构示意图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
如图1所示，在一个实施例中，提供了一种地点推荐方法的应用环境图，该应用环境图包括查询终端110和服务器120。查询终端110可通过网络与服务器120通信。查询终端110可以是智能手机、平板电脑、笔记本电脑、台式计算机中的至少一种，但并不局限于此。服务器120为位置社交网络平台(Location-based Social Network，LBSN)的服务器或者服务器集群，其数据库中存储有用户的签到数据。服务器120接收查询用户终端发送的地点推荐请求，根据携带的查询用户标识从签到数据库中查找查询用户的签到数据，然后再从签到数据库中查找与查询用户具有重叠签到地点的关联用户。服务器120根据查询用户的签到数据和关联用户的签到数据计算查询用户与关联用户的相似度，并将相似度满足设定条件的关联用户确定为查询用户对应的相似用户集中的成员。相似用户集中的成员与查询用户具有相似的地点偏好，相似用户集中成员的签到地点能够较好的贴合查询用户的偏好，且相对于查询用户的历史签到地 点具有一定的多样性。
图2为本发明一个实施例的地点推荐方法的流程示意图。应当理解的是，虽然图2的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必须按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，其可以以其他的顺序执行。而且图2中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是依次进行，而是可以与其他步骤或者其他步骤的子步骤或者阶段的至少一部分轮流或者交替执行。
参考图2，提供的地点推荐方法，该方法以应用在图1中的服务器120中进行举例说明，具体包括如下步骤：
步骤S202：接收查询用户终端发送的地点推荐请求，地点推荐请求中携带查询用户标识。
查询用户终端中安装有可与位置社交网络平台服务器通信的终端应用，查询用户终端通过该终端应用向服务器发送地点推荐请求。在一个实施例中，可通过点击应用界面中的“地点推荐”按钮的形式发送推荐请求。或者终端登录服务器后，用户通过有规则的摇动终端本体向服务器发送推荐请求。在另一个实施例中，当服务器接收到用户登录平台请求时，即视为查询用户所在终端向服务器发送了地点推荐请求，也就是，服务器为每个登录的用户进行地点推荐。
上述位置社交网络平台可以是Foursquare、Gowalla或者Facebook Places。在这些基于位置的社交网络平台中都包含了用户签到时间、签到地点以及对该地点做出的评价内容。其中，用户对地点做出的评价内容可以包括文字形式的评价，还可以包括评分评价。
步骤S204：查找与查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，签到数据集中每个用户的签到数据均包括签到地点。
服务器根据查询用户的用户标识向签到数据集中查找对应的签到数据。其中，签到数据集为根据用户在平台内的签到行为生成的签到数据的集合，用户的签到行为用户在历史时间段内于某一地点在向服务器发送的签到请求行为， 该签到请求中携带有用户对签到位置的评价信息。服务器根据用户的签到行为生成签到数据包括：获取签到时用户所处位置信息，并根据该位置信息定位签到场所，如餐厅、旅游景区等。在另一个实施例中，签到数据还包括根据用户输入的评价信息生成对签到地点的数值评分信息。
步骤S206：在签到数据集中查找查询用户的关联用户，关联用户至少有一个签到地点与查询用户的签到地点重合。
服务器查找签到数据集中是否存在与查询用户的签到地点重合的签到数据，若是，将查找到的签到数据对应的用户定义为查询用户的关联用户。关联用户与查询用户可以有一个签到地点重合，也可以有多个签到地点重合。
举例来说，查询用户u的签到数据是：a、b、c，用户v的签到数据是：a、d、e，用户w的签到数据是：b、a、f，。用户v与查询用户u有一个共同访问过的地点a，因此，用户v是查询用户u的关联用户。用户w与查询用户u有两个共同访问过的地点，分别为b和a，因此，用户w是查询用户u的关联用户。
步骤S208：计算查询用户与每个关联用户之间的相似度，并根据计算的相似度确定查询用户对应的相似用户集。
根据步骤S206确定的查询用户的关联用户可能是数量较大的用户群体，为了更精准的向查询用户推荐地点，本步骤对确定的关联用户进行筛选，从关联用户中确定与查询用户地点偏好更加相似的用户群体，也就是确定查询用户的相似用户集。
在一个实施例中，确定查询用户的相似用户集的具体方法为：根据查询用户和关联用户的签到数据计算两者之间的相似度，按照相似度由大到小的排列顺序，选择排序在前的设定数量的相似度对应的关联用户作为查询用户的相似用户集。也就是，选择前N个最大的相似度对应的关联用户作为查询用户的相似用户集。
在根据查询用户和关联用户的签到数据计算查询用户与关联用户的相似度时，可以时，关联用户与查询用户共同访问的地点越多，两者的相似度越大。或者为查询用户与关联用户访问的地点的属性越接近，两者的相似度越大。
步骤S210：确定相似用户集所对应的签到地点集合，签到地点集合中包括 相似用户集中所有关联用户所签到的签到地点。
步骤S212：去除签到地点集合中与查询用户重合的签到地点，得到推荐地点集合，将推荐地点集合中所包含的推荐地点推送至查询用户终端。
根据相似用户集中包括的所有关联用户的签到数据，确定相似用户集对应的签到地点集合。也就是，相似用户集中所有关联用户访问过的签到地点都可以在对应的签到地点集合中找到。
举例来说，相似用户集中的关联用户分别为：关联用户A：a、d、e；关联用户B：b，、a、f；和关联用户C：c、b、e，则该相似用户集对应的签到地点集合为{a，b，c，d，e，f}。
去除相似用户集对应的签到地点集合中查询用户签到过的签到地点，得到推荐地点集合。若查询用户的签到地点集合为{a，b，c)，则推荐地点集合为{d，e，f｝，将推荐地点集合中的推荐地点推送至查询用户终端。
本实施例中，查询用户的相似用户集中的用户与查询用户具有相似的地点偏好，相似用户集中的用户访问过的地点在一定概率上能够贴合查询用户的偏好。且以相似用户集对应的签到地点作为推荐地点基础，用户的多样性决定了推荐地点也更具多样性。推荐地点不限于查询用户本身访问的地点的同类型的地点，可以是在一定程度上贴合查询用户偏好的其他类型的地点。
在一个实施例中，签到数据还包括根据用户输入的评价信息生成对签到地点的数值评分信息。
举例来说，查询用户的签到数据可以包括(a，0.8)、(b，0.5)、(c，0.3)3条签到数据。其中，签到数据中的a、b、c为查询用户的签到地点，每条数据中的数值为查询用户对签到地点的评分分值。查询用户的每条签到数据对应具体的签到时间，当查询用户在不同的签到时间签到一个签到地点，将生成多条签到数据，如在时刻t1、t2查询用户均签到了地点a，则将产生两条签到数据，如(a，0.8)、(a，0.9)。
如图3所示，步骤S208：计算查询用户与每个关联用户之间的相似度，并根据计算的相似度确定查询用户对应的相似用户集，包括如下步骤：
步骤S302：计算每个关联用户与查询用户之间的相似度，其中，查询用户 与关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大。
具体的，查询用户与关联用户的签到地点是否集中判定取决于如下因素：签到相同地点的次数总和(考虑其中一个用户签到共同签到地点多次的情况)和签到非共同签到地点的次数总和。签到相同地点的次数总和越大且签到非共同签到地点的次数总和越小，则关联用户与查询用户的签到地点越集中。
举例来说，查询用户签到地点的集合分别为{a，b，c，a，d，e，f，e，h}，关联用户A签到地点集合为{a，m，i，l，m，b，f，k，h}，关联用户B签到地点集合为{a，b，e，o，f，k，m}
关联用户A与查询用户签到相同地点次数总和为5，分别为a、a、b、f、h，签到非共同签到地点的次数为9，分别为c、d、e、e、m、i、l、m、k；关联用户B与查询用户签到相同地点次数总和为6，分别为a、b、a、e、e、f，签到非共同签到地点的次数为6，分别为c、d、h、o、m、k。由上述分析可知，关联用户B与查询用户的签到地点更加集中。
在一个实施例中，可通过计算查询用户与关联用户对相同签到地点的评分的方差值或者标准差或者评分差的绝对值判定两者对共同签到地点的评分是否比较接近。以评分差的绝对值为例，如查询用户与关联用户的共同访问的地点分别为a和b，对两者的评分分别为{0.6，0.8}、{0.5，0.9}，则两者的评分接近程度为|0.6-0.5|+|0.8-0.9|，该值越小，关联用户与查询用户之间的评分越接近，相似度越大。
步骤S304：将相似度大于设定阈值的关联用户组成查询用户的相似用户集。
根据签到地点是否集中，对共同签到地点的评分是否接近这两个因素计算查询用户和关联用户之间的相似度，将相似度大于设定阈值的关联用户作为查询用户相似用户集中的成员。
其中，阈值的大小可以是预先设定的，例如可以是0.8。相似度大于0.8的关联用户组成查询用户的相似用户集。若根据预设的阈值确定的相似用户集中成员规模较小时(如确定的相似用户集中关联用户的个数小于2个)，则调整设定阈值的大小，重新确定相似用户集。
在一个实施例中，服务器预先预设多个级别的相似度阈值，如精准阈值(如 0.8)、标准阈值(如0.6)、粗放阈值(如0.4)，精准阈值＞标准阈值＞粗放阈值。当根据精准阈值无法确定相似用户集或者确定的相似用户集的规模较小时，则调整阈值为标准阈值。进一步的，若根据标准阈值无法确定相似用户集或者确定的相似用户集的规模较小时，则调整阈值为粗放阈值。
本实施例中，综合考虑签到次数和签到地点评分，可使计算出的相似度更加能够体现关联用户与查询用户之间的相似程度，即可使根据相似用户进行的地点的推荐更加能够贴和用户对地点的偏好。
在一个实施例中，可通过下述公式计算每个关联用户与查询用户之间的相似度：
Figure PCTCN2017099735-appb-000001
其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
Figure PCTCN2017099735-appb-000002
为关联用户与查询用户共同的签到地点；
Figure PCTCN2017099735-appb-000003
为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对地点i的评分；rj为查询用户或者关联用户对地点j的签到次数；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数，对于同一位置社交网络平台rmax实质上为一个定值，其作用是进行分值的归一化处理。
举例来说，假设有两个用户u和v，rmax假设为5，对应的签到数据分别为：
u：(a，0.6)、(b，0.5)、(a，0.7)
v：(a，0.5)、(c，0.4)，则公式(1)中的i签到地点a，j指代签到地点b和c，
Figure PCTCN2017099735-appb-000004
则
Figure PCTCN2017099735-appb-000005
通过公式(1)计算的相似度不仅衡量了两个用户共同访问过的地点的集合， 而且充分考虑了其他非共同访问的地点，也就是两者访问地点的分散度(或者集中度)，加之用户对签到地点的评分因素的考量，使得计算的相似度能够比较准确地评估两者之间的地点偏好是否相似。
在一个实施例中，如图4所示，步骤S210：除签到地点集合中与查询用户重合的签到地点，得到推荐地点集合，将推荐地点集合中所包含的推荐地点推送至查询用户终端的步骤包括：
步骤S402：去除签到地点集合中与查询用户重合的签到地点，得到待推荐地点集合。
相似用户集中所有查询用户未曾签到过的地点为待推荐地点集合。向查询用户推荐的地点应该是查询用户未曾访问过的地点。
本实施例中，向查询用户待推荐的地点以相似用户集对应的签到地点集合中抽取出来。相似用户集中的关联用户与查询用户具有一定的地点偏好相似度，基于相似用户集对应的签到地点进行查询用户地点推荐在一定程度上贴合了查询用户的偏好。
步骤S404：计算查询用户与待推荐地点集合中每一个待推荐地点的兴趣度，将兴趣度大于设定阈值的待推荐地点推送至查询用户终端；其中，兴趣度是通过查询用户与相似用户集中签到待推荐地点的关联用户之间的相似度以及关联用户对待推荐地点的评分计算得到的。
为了使推荐的地点更加精准的贴合查询用户的喜好，本实施例中，对确定的待推荐地点进行进一步的精准挑选，以挑选出最能够贴合查询用户真实偏好的地点。具体为：计算待推荐地点与查询用户之间的兴趣度。待推荐地点与查询用户之间的兴趣度越高，该地点与查询用户偏好之间的贴合度越高。
本实施例中，首先在相似用户集中查找签到过待推荐地点的关联用户。然后根据步骤S208中计算的查找的关联用户与查询用户之间的相似度以及关联用户对该签到地点的评分来计算查询用户与待查询地点的兴趣度。也就是，通过关联用户与查询用户之间的相似度关系以及关联用户与地点的评分关系，得到查询用户与地点的兴趣度关系。
举例来说，计算待推荐地点A与查询用户之间的兴趣度。在相似用户集中查找访问过待推荐地点A的关联用户分别为关联用户u1、u2、u3。关联用户u1、u2、u3各自的签到数据中包括有其对推荐地点A的评分信息，根据计算的关联用户与查询用户的相似度以及关联用户对推荐地点A的评分计算查询用户与推荐地点A的兴趣度，其中，关联用户对推荐地的评分越高、关联用户与查询用户的相似度越高，推荐地点与查询用户的兴趣度越高。
在一个实施例中，查询用户与待推荐地点之间的兴趣度可通过下述公式(2)来计算：
Figure PCTCN2017099735-appb-000006
其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为查询用户的相似用户集，uk是相似用户集中的签到过待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
Figure PCTCN2017099735-appb-000007
为关联用户uk对待推荐地点j的评分。
假设：计算查询用户u与待推荐地点A之间的兴趣度。在u的相似用户集中查找签到过待推荐地点A的关联用户分别为关联用户u1、u2、u3，关联用户u1、u2、u3对待推荐地点A的评分分别为
Figure PCTCN2017099735-appb-000008
查询用户与关联用户u1、u2、u3的相似度分别为sim(u，u1)、sim(u，u2)和sim(u，u3)；则查询用户u与待推荐地点A之间的兴趣度：
Figure PCTCN2017099735-appb-000009
公式(2)利用查询用户与关联用户以及关联用户与待推荐地点的关系，可计算出查询用户与待推荐地点的兴趣度，该兴趣度能够很好的体现查询用户对待推荐地点感兴趣程度。将感兴趣程度较大的待推荐地点推送至查询用户终端，使得推送的地点更加贴合用户本身真实的偏好，即实现了针对用户的精准推送。
在一个实施例中，如图5所示，提供了一种地点推荐装置，该装置包括：
请求接收模块502，用于接收查询用户终端发送的地点推荐请求，地点推荐 请求中携带查询用户标识。
签到数据查找模块504，查找与查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，签到数据集中每个用户的签到数据均包括签到地点。
关联用户确定模块506，在签到数据集中查找查询用户的关联用户，关联用户至少有一个签到地点与查询用户的签到地点重合。
相似用户集确定模块508，用于计算查询用户与每个关联用户之间的相似度，并根据计算的相似度确定查询用户对应的相似用户集。
签到地点集合确定模块510，用于确定相似用户集所对应的签到地点集合，签到地点集合中包括相似用户集中所有关联用户所签到的签到地点。
推荐地点确定模块512，用于去除签到地点集合中与查询用户重合的签到地点，得到推荐地点集合，将推荐地点集合中所包含的推荐地点推送至查询用户终端。
在一个实施例中，签到数据还包括对签到地点的评分；相似用户集确定模块508，还用于计算每个关联用户与查询用户之间的相似度，其中，查询用户与关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大；将相似度大于设定阈值的关联用户组成查询用户的相似用户集。
在一个实施例中，通过下述公式计算每个关联用户与查询用户之间的相似度：
Figure PCTCN2017099735-appb-000010
其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
Figure PCTCN2017099735-appb-000011
为关联用户与查询用户共同的签到地点；
Figure PCTCN2017099735-appb-000012
为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对地点i的评分；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数。
在一个实施例中，推荐地点确定模块512，还用于去除签到地点集合中与查询用户重合的签到地点，得到待推荐地点集合；计算查询用户与待推荐地点集合中每一个待推荐地点的兴趣度，将兴趣度大于设定阈值的待推荐地点推送至查询用户终端，其中，其中，兴趣度是通过查询用户与相似用户集中签到待推荐地点的关联用户之间的相似度以及关联用户对待推荐地点的评分计算得到的。
在一个实施例中，查询用户与待推荐地点的兴趣度的计算公式为：
Figure PCTCN2017099735-appb-000013
其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为查询用户的相似用户集，uk是相似用户集中的签到过待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
Figure PCTCN2017099735-appb-000014
为关联用户uk对待推荐地点j的评分。
在一个实施例中，上述各个实施例中的数据发布装置可以实现为一种计算程序的形式，计算机程序对应的计算机可执行指令可在如图6所示的计算机设备上运行。该计算机设备可以是一台物理服务器，也可以是多台服务器构成的服务器集群。其内部结构为：包括通过系统总线连接的处理器、非易失性存储介质、内存储器和网络接口。其中，该计算机设备的非易失性存储介质存储有操作系统、数据库和至少一条上述的由数据发布装置实现的计算机可执行指令。数据库用于存储数据，如存储用户的签到数据。处理器用于提供计算和控制能力，支撑整个计算机设备的运行。内存储器为非易失性存储介质中的操作系统和用于实现数据发布的计算机可执行指令的运行提供环境。网络接口用于与查询终端进行通信连接。
本领域技术人员可以理解，图6中示出的计算机设备的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的服务器的限定，具体的服务器可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
上述的网络接口可以是以太网卡或者无线网卡等。上述各模块还可以硬件形式内嵌于或者独立于上述的计算机设备中。也可以如上述的以软件的形式存储于差分发布服务器的存储器中，以便于处理器调用执行以上各个模块对应的操作。该处理器可以为中央处理单元(CPU)、微处理器、单片机等。
在一个实施例中，提供了一个或多个存储有计算机可执行指令的非易失性可读存储介质，该指令被一个或多个处理器执行时，使得一个或多个处理器执行上述各实施例方法中的全部或部分流程。上述的计算机可执行指令为由上述各实施例方法中的全部或者部分流程实现的计算机程序对应的计算机可执行指令。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机可执行指令来指令相关的硬件来完成，程序可存储于一计算机可读取存储介质中，如本发明实施例中，该程序可存储于计算机系统的非易失性可读存储介质中，并被该计算机系统中的至少一个处理器执行，以实现包括如上述各方法的实施例的流程。其中，非易失性可读存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory，ROM)等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。

Claims (20)

  1. 一种地点推荐方法，所述方法包括：
    接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
    查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
    在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
    计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
    确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；及
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
  2. 根据权利要求1所述的方法，其特征在于，所述签到数据还包括对所述签到地点的评分；
    所述计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集，包括：
    计算每个所述关联用户与所述查询用户之间的相似度，其中，所述查询用户与所述关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大；
    将所述相似度大于设定阈值的所述关联用户组成所述查询用户的相似用户集。
  3. 根据权利要求2所述的方法，其特征在于，所述计算每个所述关联用户与所述查询用户之间的相似度为：
    通过下述公式计算每个所述关联用户与所述查询用户之间的相似度：
    Figure PCTCN2017099735-appb-100001
    其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
    Figure PCTCN2017099735-appb-100002
    为关联用户与查询用户共同的签到地点；
    Figure PCTCN2017099735-appb-100003
    为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对地点i的评分；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数。
  4. 根据权利要求2所述的方法，其特征在于，所述去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端，包括：
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到待推荐地点集合；
    计算所述查询用户与所述待推荐地点集合中每一个待推荐地点的兴趣度，将所述兴趣度大于设定阈值的所述待推荐地点推送至所述查询用户终端；
    其中，所述兴趣度是通过所述查询用户与所述相似用户集中签到所述待推荐地点的关联用户之间的所述相似度以及所述关联用户对所述待推荐地点的评分计算得到的。
  5. 根据权利要求4所述的方法，其特征在于，所述查询用户与所述待推荐地点的兴趣度的计算公式为：
    Figure PCTCN2017099735-appb-100004
    其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为所述查询用户的相似用户集，uk是所述相似用户集中的签到过所述待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
    Figure PCTCN2017099735-appb-100005
    为关联用户uk对待推荐地点j的评分。
  6. 一种地点推荐装置，其特征在于，所述装置包括：
    请求接收模块，用于接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
    签到数据查找模块，用于查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
    关联用户确定模块，用于在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
    相似用户集确定模块，用于计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
    签到地点集合确定模块，用于确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；
    推荐地点确定模块，用于去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
  7. 根据权利要求6所述的装置，其特征在于，所述签到数据还包括对所述签到地点的评分；所述相似用户集确定模块，还用于计算每个所述关联用户与所述查询用户之间的相似度，其中，所述查询用户与所述关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大；将所述相似度大于设定阈值的所述关联用户组成所述查询用户的相似用户集。
  8. 根据权利要求7所述的装置，其特征在于，所述相似用户集确定模块，还用于通过下述公式计算每个所述关联用户与所述查询用户之间的相似度：
    Figure PCTCN2017099735-appb-100006
    其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
    Figure PCTCN2017099735-appb-100007
    为关联用户与查询用户共同的签到地点；
    Figure PCTCN2017099735-appb-100008
    为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对 地点i的评分；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数。
  9. 根据权利要求7所述的装置，其特征在于，所述推荐地点确定模块，还用于去除所述签到地点集合中与所述查询用户重合的签到地点，得到待推荐地点集合；计算所述查询用户与所述待推荐地点集合中每一个待推荐地点的兴趣度，将所述兴趣度大于设定阈值的所述待推荐地点推送至所述查询用户终端；
    其中，所述兴趣度是通过所述查询用户与所述相似用户集中签到所述待推荐地点的关联用户之间的所述相似度以及所述关联用户对所述待推荐地点的评分计算得到的。
  10. 根据权利要求9所述的装置，其特征在于，所述查询用户与所述待推荐地点的兴趣度的计算公式为：
    Figure PCTCN2017099735-appb-100009
    其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为所述查询用户的相似用户集，uk是所述相似用户集中的签到过所述待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
    Figure PCTCN2017099735-appb-100010
    为关联用户uk对待推荐地点j的评分。
  11. 一种计算机设备，包括存储器和处理器，所述存储器中存储有计算机可执行指令，所述指令被所述处理器执行时，使得所述处理器执行以下步骤：
    接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
    查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
    在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有一个签到地点与所述查询用户的签到地点重合；
    计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
    确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；及
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
  12. 根据权利要求11所述的计算机设备，其特征在于，所述签到数据还包括对所述签到地点的评分；
    所述处理器执行的所述计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集，包括：
    计算每个所述关联用户与所述查询用户之间的相似度，其中，所述查询用户与所述关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大；
    将所述相似度大于设定阈值的所述关联用户组成所述查询用户的相似用户集。
  13. 根据权利要求12所述的计算机设备，其特征在于，所述处理器执行的所述计算每个所述关联用户与所述查询用户之间的相似度为：
    通过下述公式计算每个所述关联用户与所述查询用户之间的相似度：
    Figure PCTCN2017099735-appb-100011
    其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
    Figure PCTCN2017099735-appb-100012
    为关联用户与查询用户共同的签到地点；
    Figure PCTCN2017099735-appb-100013
    为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对地点i的评分；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数。
  14. 根据权利要求12所述的计算机设备，其特征在于，所述处理器执行的所述去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点 集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端，包括：
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到待推荐地点集合；
    计算所述查询用户与所述待推荐地点集合中每一个待推荐地点的兴趣度，将所述兴趣度大于设定阈值的所述待推荐地点推送至所述查询用户终端；
    其中，所述兴趣度是通过所述查询用户与所述相似用户集中签到所述待推荐地点的关联用户之间的所述相似度以及所述关联用户对所述待推荐地点的评分计算得到的。
  15. 根据权利要求14所述的计算机设备，其特征在于，所述查询用户与所述待推荐地点的兴趣度的计算公式为：
    Figure PCTCN2017099735-appb-100014
    其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为所述查询用户的相似用户集，uk是所述相似用户集中的签到过所述待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
    Figure PCTCN2017099735-appb-100015
    为关联用户uk对待推荐地点j的评分。
  16. 一个或者多个存储有计算机可执行指令的非易失性可读存储介质，所述指令被一个或者多个处理器执行，使得所述一个或者多个处理器执行以下步骤：
    接收查询用户终端发送的地点推荐请求，所述地点推荐请求中携带查询用户标识；
    查找与所述查询用户标识对应的查询用户的签到数据，其中，位置社交网络平台根据用户的历史签到行为生成签到数据集，所述签到数据集中每个用户的签到数据均包括签到地点；
    在所述签到数据集中查找所述查询用户的关联用户，所述关联用户至少有 一个签到地点与所述查询用户的签到地点重合；
    计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集；
    确定所述相似用户集所对应的签到地点集合，所述签到地点集合中包括所述相似用户集中所有关联用户所签到的签到地点；及
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端。
  17. 根据权利要求16所述的非易失性可读存储介质，其特征在于，所述签到数据还包括对所述签到地点的评分；
    所述处理器执行的所述计算查询用户与每个所述关联用户之间的相似度，并根据计算的所述相似度确定所述查询用户对应的相似用户集，包括：
    计算每个所述关联用户与所述查询用户之间的相似度，其中，所述查询用户与所述关联用户所签到的签到地点越集中、对共同签到地点的评分越接近，计算得到的相似度数值越大；
    将所述相似度大于设定阈值的所述关联用户组成所述查询用户的相似用户集。
  18. 根据权利要求17所述的非易失性可读存储介质，其特征在于，所述处理器执行的所述计算每个所述关联用户与所述查询用户之间的相似度为：
    通过下述公式计算每个所述关联用户与所述查询用户之间的相似度：
    Figure PCTCN2017099735-appb-100016
    其中，u、v分别代表查询用户和关联用户；sim(u，v)为关联用户与查询用户之间的相似度；
    Figure PCTCN2017099735-appb-100017
    为关联用户与查询用户共同的签到地点；
    Figure PCTCN2017099735-appb-100018
    为关联用户与查询用户非共同的签到地点；Rui和Rvi分别为查询用户、关联用户对地点i的评分；rmax为位置社交网络平台中被任一用户签到次数最多的签到地点对应的签到次数。
  19. 根据权利要求17所述的非易失性可读存储介质，其特征在于，所述处 理器执行的所述去除所述签到地点集合中与所述查询用户重合的签到地点，得到推荐地点集合，将所述推荐地点集合中所包含的推荐地点推送至所述查询用户终端，包括：
    去除所述签到地点集合中与所述查询用户重合的签到地点，得到待推荐地点集合；
    计算所述查询用户与所述待推荐地点集合中每一个待推荐地点的兴趣度，将所述兴趣度大于设定阈值的所述待推荐地点推送至所述查询用户终端；
    其中，所述兴趣度是通过所述查询用户与所述相似用户集中签到所述待推荐地点的关联用户之间的所述相似度以及所述关联用户对所述待推荐地点的评分计算得到的。
  20. 根据权利要求19所述的非易失性可读存储介质，其特征在于，所述查询用户与所述待推荐地点的兴趣度的计算公式为：
    Figure PCTCN2017099735-appb-100019
    其中，u为查询用户，j为确定的待推荐地点集合中的待推荐地点；U为所述查询用户的相似用户集，uk是所述相似用户集中的签到过所述待推荐地点j的关联用户；sim(u，uk)为查询用户u与关联用户uk之间的相似度，
    Figure PCTCN2017099735-appb-100020
    为关联用户uk对待推荐地点j的评分。
PCT/CN2017/099735 2017-06-12 2017-08-30 地点推荐方法、装置、计算机设备和存储介质 Ceased WO2018227773A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710439338.0 2017-06-12
CN201710439338.0A CN107679053B (zh) 2017-06-12 2017-06-12 地点推荐方法、装置、计算机设备及存储介质

Publications (1)

Publication Number Publication Date
WO2018227773A1 true WO2018227773A1 (zh) 2018-12-20

Family

ID=61133573

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/099735 Ceased WO2018227773A1 (zh) 2017-06-12 2017-08-30 地点推荐方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN107679053B (zh)
WO (1) WO2018227773A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112685525A (zh) * 2020-12-29 2021-04-20 北京三快在线科技有限公司 位置推荐方法、装置、设备以及存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447731B (zh) * 2018-09-18 2024-10-18 平安科技（深圳）有限公司 跨平台产品推荐方法、装置、计算机设备和存储介质
CN110059248B (zh) * 2019-03-21 2022-12-13 腾讯科技（深圳）有限公司 一种推荐方法、装置及服务器
CN110866180B (zh) * 2019-10-12 2022-07-29 平安国际智慧城市科技股份有限公司 资源推荐方法、服务器及存储介质
CN111523031B (zh) * 2020-04-22 2023-03-31 北京百度网讯科技有限公司 用于推荐兴趣点的方法和装置
CN111737537B (zh) * 2020-07-21 2020-11-27 杭州欧若数网科技有限公司 基于图数据库的poi推荐方法、设备及介质

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105740401A (zh) * 2016-01-28 2016-07-06 北京理工大学 一种基于个体行为和群体兴趣的兴趣地点推荐方法及装置
CN106056455A (zh) * 2016-06-02 2016-10-26 南京邮电大学 一种基于位置和社交关系的群组与地点推荐方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103488678B (zh) * 2013-08-05 2017-02-08 北京航空航天大学 一种基于用户签到相似度的好友推荐系统
CN104978437B (zh) * 2015-07-22 2018-09-25 浙江大学 一种基于地理位置的推荐方法和推荐系统

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105740401A (zh) * 2016-01-28 2016-07-06 北京理工大学 一种基于个体行为和群体兴趣的兴趣地点推荐方法及装置
CN106056455A (zh) * 2016-06-02 2016-10-26 南京邮电大学 一种基于位置和社交关系的群组与地点推荐方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112685525A (zh) * 2020-12-29 2021-04-20 北京三快在线科技有限公司 位置推荐方法、装置、设备以及存储介质

Also Published As

Publication number Publication date
CN107679053A (zh) 2018-02-09
CN107679053B (zh) 2020-02-18

Similar Documents

Publication Publication Date Title
US11294981B2 (en) System and method for large scale crowdsourcing of map data cleanup and correction
WO2018227773A1 (zh) 地点推荐方法、装置、计算机设备和存储介质
US9965522B2 (en) Mobile application search ranking
CN107766462B (zh) 基于用户偏好、社交信誉度和地理位置的兴趣点推荐方法
CN103488766B (zh) 应用程序搜索方法和装置
CN104937590B (zh) 用相关第三方应用内容扩充搜索结果
RU2696230C2 (ru) Поиск, основанный на комбинировании пользовательских данных отношений
CN104281882B (zh) 基于用户特征的预测社交网络信息流行度的方法及系统
CN107480714B (zh) 基于全视角特征的跨社交网络用户识别方法
JP6608972B2 (ja) ソーシャルネットワークに基づいてグループを探索する方法、デバイス、サーバ及び記憶媒体
WO2018077181A1 (zh) 图中心性计算方法、装置及存储介质
US11030265B2 (en) Cross-platform data matching method and apparatus, computer device and storage medium
CN111177125A (zh) 用于分析关注实体特性的设备、系统以及方法
US9159030B1 (en) Refining location detection from a query stream
US9946799B2 (en) Federated search page construction based on machine learning
CN105677765A (zh) 给用户推荐期望的功能序列的方法及系统
CN110213164B (zh) 一种基于拓扑信息融合的识别网络关键传播者的方法及装置
US20160034583A1 (en) Adjusting search results based on overlapping work histories
US20140025673A1 (en) Techniques for estimating distance between members of a social network service
WO2019205373A1 (zh) 相似用户查找装置、方法及计算机可读存储介质
CN107145545A (zh) 一种基于位置的社交网络中Top‑k区域用户文本数据推荐方法
CN113434762A (zh) 基于用户信息的关联推送方法、装置、设备及存储介质
WO2020047921A1 (zh) 基于层次三元组损失函数的深度度量学习方法及其装置
EP3100230A1 (en) Segments of contacts
Bae et al. AppTrends: A graph-based mobile app recommendation system using usage history

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17913491

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17913491

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 08.04.2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17913491

Country of ref document: EP

Kind code of ref document: A1