CN110968771B - Job recommendation cold start method and system based on friendship relationship - Google Patents

Job recommendation cold start method and system based on friendship relationship Download PDF

Info

Publication number
CN110968771B
CN110968771B CN201811145712.7A CN201811145712A CN110968771B CN 110968771 B CN110968771 B CN 110968771B CN 201811145712 A CN201811145712 A CN 201811145712A CN 110968771 B CN110968771 B CN 110968771B
Authority
CN
China
Prior art keywords
user
information
friend
job
representation
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
Application number
CN201811145712.7A
Other languages
Chinese (zh)
Other versions
CN110968771A (en
Inventor
周华
牛章鹏
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.)
Beijing Taoyoutianxia Technology Co ltd
Original Assignee
Beijing Taoyoutianxia Technology 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 Beijing Taoyoutianxia Technology Co ltd filed Critical Beijing Taoyoutianxia Technology Co ltd
Priority to CN201811145712.7A priority Critical patent/CN110968771B/en
Publication of CN110968771A publication Critical patent/CN110968771A/en
Application granted granted Critical
Publication of CN110968771B publication Critical patent/CN110968771B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06Q10/10—Office automation; Time management
    • G06Q10/105—Human resources

Landscapes

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

Abstract

The invention discloses a job position recommending cold start method and system based on friend relation, comprising the following steps: A. acquiring a friend relationship of a user; B. constructing a representation of user i using feature vectorsA representation; C. constructing a representation of position z using feature vectorsA representation; D. calculating a correlation score of the user and the position according to the bilinear model; E. ranking the positions according to the relevance scores to finish personalized recommendation, and the invention can calculate the relevance scores of the positions and the users according to the user information and the friend information and recommend the most relevant positions to the users according to the relevance scores; when the user information is incomplete, the relevance score can be calculated according to the friend information of the user, the dynamic information of the position and the bilinear model, the position is recommended, and the cold start problem of position recommendation is solved.

Description

Job recommendation cold start method and system based on friendship relationship
Technical Field
The invention relates to the technical field of recruitment, in particular to a job recommendation cold start method and system based on friend relations.
Background
Currently, in recruitment industry, a job recommendation system needs to acquire a great amount of detailed information of a user, such as industry, direction, job position, skills, compensation range, job site, and the like, then match the information with job position information (industry, direction, job position, skill requirement, compensation, job site, and the like), and recommend a job position with higher matching degree to the user.
Under the condition of cold start, namely when the user is a new user or the user information is partially or completely lost, the recommendation system cannot accurately match positions, and can only recommend new positions or hot positions to the user, so that most of the identical position information is displayed to a large number of new users or information-lost users, individuation cannot be achieved, and the requirements of different users cannot be met.
Disclosure of Invention
The invention aims to provide a job recommendation cold start method and system based on friend relations, which are used for solving the problems in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions: a job recommendation cold start method based on friend relation comprises the following steps:
A. acquiring a friend relationship of a user: when a user registers, the address book needs to be uploaded, and a topological relation between the users is constructed according to the address book, so that a friend set of the user is obtained;
B. constructing a representation of user i using feature vectors A representation;
C. Constructing a representation of position z using feature vectors The representation is: wherein, the characteristics of position include: static information of job position: industry, direction, skill, age, job site, salary range; dynamic information of job position: freshness, click rate;
D. according to the bilinear model, calculating a correlation score of the user and the position: based on the bilinear model, the interest score of user i for position j is:
Wherein: u i,m: the mth element in user feature vector u i
Z j,n: nth element in position feature vector z j
W mn: weights for measuring the interrelationship of u i,m and z j,n
W: a weight matrix of size m×n;
E. And ordering the positions according to the relevance scores to finish personalized recommendation.
Preferably, the features of the user in the step B include: demographic information of the user: industry, direction, skill, age, gender, job site, salary range; statistics of behavior information of users: number of logins per week/month/year, number of browses/clicks at various locations of the website; statistical histograms of demographic information of friends; statistical histogram of behavior information of friends, normalizing histogram of demographic information of friends, adding the normalized histogram into feature vector, and combining all feature vectors to form feature vector
Preferably, the weight parameter W of the bilinear model in the step D is learned from historical data, and the calculation steps are as follows:
a. Historical training data For the corresponding set of show/click events r ij e { -1, 1|;
b. using a logistic function as a likelihood function:
the likelihood of the training data is:
c. further assume that the prior probability of W satisfies the gaussian distribution:
d. according to the bayesian theorem, this problem is converted into a problem of maximum posterior probability. The logarithm can be taken and converted into the following optimization problem:
w in the equation is solved using an optimization numerical algorithm.
Preferably, the job recommendation cold start system based on friend relation comprises a user information acquisition module for acquiring user information and friend information; the correlation score calculation module is used for calculating the correlation score of the position and the user according to the user information and the friend information; and the position recommending module is used for recommending positions to the user according to the relevance scores.
Compared with the prior art, the invention has the beneficial effects that: according to the invention, the correlation score between the position and the user can be calculated according to the user information and the friend information, and the most relevant position is recommended to the user according to the correlation score; when the user information is incomplete, the relevance score can be calculated according to the friend information of the user, the dynamic information of the position and the bilinear model, the position is recommended, and the cold start problem of position recommendation is solved.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the present invention provides a technical solution: a job recommendation cold start method based on friend relation comprises the following steps:
A. acquiring a friend relationship of a user: when a user registers, the address book needs to be uploaded, and a topological relation between the users is constructed according to the address book, so that a friend set of the user is obtained;
B. constructing a representation of user i using feature vectors A representation;
C. Constructing a representation of position z using feature vectors The representation is: wherein, the characteristics of position include: static information of job position: industry, direction, skill, age, job site, salary range; dynamic information of job position: freshness, click rate;
D. according to the bilinear model, calculating a correlation score of the user and the position: based on the bilinear model, the interest score of user i for position j is:
Wherein: u i,m: the mth element in user feature vector u i
Z j,n: nth element in position feature vector z j
W mn: weights for measuring the interrelationship of u i,m and z j,n
W: a weight matrix of size m×n;
E. And ordering the positions according to the relevance scores to finish personalized recommendation.
In the present invention, the features of the user in step B include: demographic information of the user: industry, direction, skill, age, gender, job site, salary range; statistics of behavior information of users: number of logins per week/month/year, number of browses/clicks at various locations of the website; statistical histograms of demographic information of friends; statistical histograms of behavior information of friends are added into feature vectors after normalization of histograms of demographic information of friends, wherein the demographic information is category features and is represented by binary vectors. For example, sex "man" is denoted as [0,1], sex "woman" is denoted as [1,0], and sex information is denoted as [0,0] if the sex information is missing, so that the information missing does not affect the bilinear model, all feature vectors are combined to form a feature vector
In the invention, the weight parameter W of the bilinear model in the step D is learned by historical data, and the calculation steps are as follows:
a. Historical training data For the set of corresponding show/click events r ij e { -1,1 };
b. using a logistic function as a likelihood function:
the likelihood of the training data is:
c. further assume that the prior probability of W satisfies the gaussian distribution:
d. according to the bayesian theorem, this problem is converted into a problem of maximum posterior probability. The logarithm can be taken and converted into the following optimization problem:
w in the equation is solved using an optimization numerical algorithm.
In addition, the invention also discloses a post recommendation cold start system based on friend relation, which comprises a user information acquisition module for acquiring user information and friend information; the correlation score calculation module is used for calculating the correlation score of the position and the user according to the user information and the friend information; and the position recommending module is used for recommending positions to the user according to the relevance scores.
In summary, the invention can calculate the relevance score of the position and the user according to the user information and the friend information, and recommend the most relevant position to the user according to the relevance score; when the user information is incomplete, the relevance score can be calculated according to the friend information of the user, the dynamic information of the position and the bilinear model, the position is recommended, and the cold start problem of position recommendation is solved.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (2)

1. A job recommendation cold start method based on friend relation is characterized by comprising the following steps: the method comprises the following steps:
A. acquiring a friend relationship of a user: when a user registers, the address book needs to be uploaded, and a topological relation between the users is constructed according to the address book, so that a friend set of the user is obtained;
B. constructing a representation of a user u, and representing the representation by a feature vector u i∈RM;
C. Building a representation of position z, represented by feature vector z j∈RN: wherein, the characteristics of position include: static information of job position: industry, direction, skill, age, job site, salary range; dynamic information of job position: freshness, click rate;
D. according to the bilinear model, calculating a correlation score of the user and the position: based on the bilinear model, the interest score of user i for position j is:
wherein: u i,m: the m-th element in the user feature vector u i;
z j,n: the nth element in the position feature vector z j;
W mn: weights for measuring the interrelationship of u i,m and z j,n;
w: a weight matrix of size m×n;
E. ranking the positions according to the relevance scores to finish personalized recommendation;
And D, learning the weight parameter W of the bilinear model from historical data, wherein the calculation steps are as follows:
a. Historical training data O is a set of corresponding display and click events r ij epsilon-1, 1;
b. using a logistic function as a likelihood function:
the likelihood of the training data is:
c. further assume that the prior probability of W satisfies the gaussian distribution:
d. According to the Bayes theorem, the problem is converted into a problem for solving the maximum posterior probability, and the logarithm is taken to be converted into the following optimization problem:
w in the equation is solved using an optimization numerical algorithm.
2. The job recommendation cold start method based on friend relation according to claim 1, wherein: the features of the user in the step B include: demographic information of the user: industry, direction, skill, age, gender, job site, salary range; statistics of behavior information of users: number of logins per week/month/year, number of browses/clicks at various locations of the website; statistical histograms of demographic information of friends; and (3) normalizing the statistical histogram of the behavior information of the friend, adding the normalized histogram of the demographic information of the friend into the feature vector, and combining all the feature vectors to form a feature vector u i∈RM.
CN201811145712.7A 2018-09-29 2018-09-29 Job recommendation cold start method and system based on friendship relationship Active CN110968771B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811145712.7A CN110968771B (en) 2018-09-29 2018-09-29 Job recommendation cold start method and system based on friendship relationship

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811145712.7A CN110968771B (en) 2018-09-29 2018-09-29 Job recommendation cold start method and system based on friendship relationship

Publications (2)

Publication Number Publication Date
CN110968771A CN110968771A (en) 2020-04-07
CN110968771B true CN110968771B (en) 2024-05-28

Family

ID=70027216

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811145712.7A Active CN110968771B (en) 2018-09-29 2018-09-29 Job recommendation cold start method and system based on friendship relationship

Country Status (1)

Country Link
CN (1) CN110968771B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112115364A (en) * 2020-09-23 2020-12-22 北京明略昭辉科技有限公司 Recommendation method and system for cold start based on login operation and computer equipment

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102317941A (en) * 2011-07-30 2012-01-11 华为技术有限公司 Information recommending method, recommending engine and network system
CN102426686A (en) * 2011-09-29 2012-04-25 南京大学 Internet information product recommending method based on matrix decomposition
CN103544312A (en) * 2013-11-04 2014-01-29 成都数之联科技有限公司 Employment information matching method based on social network
CN105447015A (en) * 2014-08-18 2016-03-30 上海众聘信息科技有限公司 Mobile terminal based job information pushing method, apparatus and system
CN106022723A (en) * 2016-05-19 2016-10-12 中国传媒大学 Personalized recommendation method of employment information
CN106886559A (en) * 2016-12-28 2017-06-23 浙江浙大网新集团有限公司 The collaborative filtering method of good friend's feature and similar users feature is incorporated simultaneously
CN106933821A (en) * 2015-12-29 2017-07-07 中国电信股份有限公司 A kind of personalized position based on Similarity Measure recommends method and system
CN106971345A (en) * 2016-01-08 2017-07-21 车海莺 A kind of location recommendation method based on position social networks
CN107220328A (en) * 2017-05-23 2017-09-29 南京大学 The video recommendation method of weak relation and strong relation based on social networks
CN107633100A (en) * 2017-10-23 2018-01-26 苏州大学 A kind of point of interest based on incorporation model recommends method and device

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102317941A (en) * 2011-07-30 2012-01-11 华为技术有限公司 Information recommending method, recommending engine and network system
CN102426686A (en) * 2011-09-29 2012-04-25 南京大学 Internet information product recommending method based on matrix decomposition
CN103544312A (en) * 2013-11-04 2014-01-29 成都数之联科技有限公司 Employment information matching method based on social network
CN105447015A (en) * 2014-08-18 2016-03-30 上海众聘信息科技有限公司 Mobile terminal based job information pushing method, apparatus and system
CN106933821A (en) * 2015-12-29 2017-07-07 中国电信股份有限公司 A kind of personalized position based on Similarity Measure recommends method and system
CN106971345A (en) * 2016-01-08 2017-07-21 车海莺 A kind of location recommendation method based on position social networks
CN106022723A (en) * 2016-05-19 2016-10-12 中国传媒大学 Personalized recommendation method of employment information
CN106886559A (en) * 2016-12-28 2017-06-23 浙江浙大网新集团有限公司 The collaborative filtering method of good friend's feature and similar users feature is incorporated simultaneously
CN107220328A (en) * 2017-05-23 2017-09-29 南京大学 The video recommendation method of weak relation and strong relation based on social networks
CN107633100A (en) * 2017-10-23 2018-01-26 苏州大学 A kind of point of interest based on incorporation model recommends method and device

Also Published As

Publication number Publication date
CN110968771A (en) 2020-04-07

Similar Documents

Publication Publication Date Title
US11586941B2 (en) Recommendation method and apparatus
CN113508378B (en) Training method, recommendation method, device and computer readable medium for recommendation model
US10509791B2 (en) Statistical feature engineering of user attributes
US20170053297A1 (en) Enhanced systems, processes, and user interfaces for scoring assets associated with a population of data
US11580447B1 (en) Shared per content provider prediction models
JP6854748B2 (en) Information providing equipment, information providing method, and program
US20220405607A1 (en) Method for obtaining user portrait and related apparatus
US20210366006A1 (en) Ranking of business object
US9665551B2 (en) Leveraging annotation bias to improve annotations
US20220398500A1 (en) Partially local federated learning
US11651255B2 (en) Method and apparatus for object preference prediction, and computer readable medium
US20180330390A1 (en) Enhanced systems, processes, and user interfaces for targeted marketing associated with a population of assets
US20240177006A1 (en) Data processing method and apparatus, program product, computer device, and medium
George et al. Asia-Pacific demand for military expenditure: spatial panel and SUR estimates
CN109801101A (en) Label determines method, apparatus, computer equipment and storage medium
CN114565407A (en) Advertisement delivery data analysis method and system
US11823217B2 (en) Advanced segmentation with superior conversion potential
CN111161009B (en) Information pushing method, device, computer equipment and storage medium
Palm et al. Bootstrap-based inferential improvements in beta autoregressive moving average model
CN110263973B (en) Method and device for predicting user behavior
CN110968771B (en) Job recommendation cold start method and system based on friendship relationship
CN111445280A (en) Model generation method, restaurant ranking method, system, device and medium
Demirhan A generalized Gumbel distribution and its parameter estimation
CN108960954B (en) Content recommendation method and system based on user group behavior feedback
US11144953B1 (en) Determining performance metrics for delivery of electronic media content items by online publishers scaled using a baseline conversion rate

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