CN113076478A - Technical resource and service recommendation system based on hybrid recommendation algorithm - Google Patents

Technical resource and service recommendation system based on hybrid recommendation algorithm Download PDF

Info

Publication number
CN113076478A
CN113076478A CN202110399613.7A CN202110399613A CN113076478A CN 113076478 A CN113076478 A CN 113076478A CN 202110399613 A CN202110399613 A CN 202110399613A CN 113076478 A CN113076478 A CN 113076478A
Authority
CN
China
Prior art keywords
user
enterprise
module
service
resource
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.)
Granted
Application number
CN202110399613.7A
Other languages
Chinese (zh)
Other versions
CN113076478B (en
Inventor
洪佳升
贾宁
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.)
Tongji University
Original Assignee
Tongji University
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 Tongji University filed Critical Tongji University
Priority to CN202110399613.7A priority Critical patent/CN113076478B/en
Publication of CN113076478A publication Critical patent/CN113076478A/en
Application granted granted Critical
Publication of CN113076478B publication Critical patent/CN113076478B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • 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
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明基于网络协同制造技术资源服务平台提供了一种基于混合推荐算法的技术资源及服务推荐系统，包括数据预处理模块、用户语义描述模块、个性化技术资源推荐模块、企业相似度计算模块、个性化服务推荐模块、技术资源及服务潜在组合挖掘模块。所述数据预处理模块包括对用户浏览与购买记录、企业注册信息等数据的预处理；所述用户语义描述模块对用户可能感兴趣的领域信息进行建模；所述个性化技术资源推荐模块向用户推荐技术资源；所述企业相似度计算模块包括两种相似度计算方法。本发明充分利用平台数据，通过多种数据分析算法混合，全方位向用户推荐技术资源及企业服务，达到提高网络协同制造程度的目的。

Figure 202110399613

The present invention provides a technology resource and service recommendation system based on a hybrid recommendation algorithm based on a network collaborative manufacturing technology resource service platform, including a data preprocessing module, a user semantic description module, a personalized technology resource recommendation module, an enterprise similarity calculation module, Personalized service recommendation module, technical resource and service potential combination mining module. The data preprocessing module includes the preprocessing of data such as user browsing and purchase records, enterprise registration information, etc.; the user semantic description module models the domain information that the user may be interested in; the personalized technical resource recommendation module The user recommends technical resources; the enterprise similarity calculation module includes two similarity calculation methods. The invention makes full use of platform data, and through the mixing of multiple data analysis algorithms, comprehensively recommends technical resources and enterprise services to users, and achieves the purpose of improving the degree of network collaborative manufacturing.

Figure 202110399613

Description

Technical resource and service recommendation system based on hybrid recommendation algorithm
Technical Field
The invention relates to the field of online services, in particular to a technical resource and service recommendation system based on a hybrid recommendation algorithm.
Background
With the development of communication technology, information technology, cloud computing and other technologies, the enterprise manufacturing model is changed greatly. A network collaborative manufacturing model with characteristics of agile manufacturing, information sharing, best utilization of resources, and the like, has gained more and more attention and applications.
The foundation for realizing the network collaborative manufacturing mode is a network collaborative manufacturing technology resource service platform. On the platform, a large number of technical resources uploaded by users or enterprises exist, including literature resources, article resources, document resources, software resources, manufacturing resources, talent resources and the like. There are also services released by enterprises for users to purchase. The user can look up technical resources and purchase related enterprise services according to the actual project requirements, and therefore project implementation progress is accelerated.
Due to the fact that the technical resources and services on the platform are large in types and quantity, users cannot obtain interested technical resources and services at the first time. In order to increase the user experience, a platform needs to develop a set of recommendation system to recommend technical resources and services to the user.
Disclosure of Invention
The invention provides a technical resource and service recommendation system based on a hybrid recommendation algorithm, which is used for recommending technical resources and services released on a platform.
Drawings
Fig. 1 is a system structure block diagram of a technical resource and service recommendation system based on a hybrid recommendation algorithm according to the present invention.
FIG. 2 is a flow diagram of a technology resource and service potential portfolio mining module.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, a technology resource and service recommendation system based on a hybrid recommendation algorithm includes a data preprocessing module, a user semantic description module, a personalized technology resource recommendation module, an enterprise similarity calculation module, a personalized service recommendation module, and a technology resource and service potential combination mining module.
Each module is specifically described below in detail.
The data preprocessing module takes technical resource records browsed by a user, enterprise service records purchased by the user and enterprise registration information as input. Wherein, technical resource records are browsed by a user, enterprise service records purchased by the user are extracted from a server log file through a regular expression; the enterprise registration information is obtained from a database. And then clustering by using a density-based clustering algorithm DBSCAN according to the time generated by the user behavior record to obtain a user behavior cluster.
Wherein:
the user browsing technical resource records and the enterprise information vectors are respectively used for the user semantic description module and the enterprise similarity calculation module, and the obtained user browsing technical resource record cluster and the user purchasing enterprise service record cluster are provided for the technical resource and service potential combination mining module.
The data preprocessing module firstly reads in technical resource records browsed by a user, enterprise service records purchased by the user and enterprise registration information, wherein the technical resource records browsed by the user comprise information such as user ID, technical resource ID, recording time and technical resource browsing amount; the user purchase enterprise service record includes information such as a user ID, a service ID, a recording time, and the like.
According to the project requirements of users in a period of time or the assumption that interested contents are concentrated in a certain limited field, the records are clustered based on the density according to the time to obtain a cluster.
The enterprise registration information includes the enterprise-oriented domain, the registered fund of the enterprise, the number of employees of the enterprise, the establishment time of the enterprise, the business turnover, the profit margin of the enterprise, and the like.
One-hot coding is carried out on characteristics such as enterprise-oriented fields, and characteristics such as registered fund of enterprises, number of employees of enterprises, establishment time of enterprises, business turnover and profit rate of enterprises are standardized to obtain enterprise information vectors.
When the user semantic description module calculates the weight of a certain technical resource, a certain punishment is carried out according to the browsing amount of the technical resource and is used for providing the user semantic description module for the personalized technical resource recommendation module. Because many users browse popular technical resources only because of the popular technical resources, the individuality of the users cannot be reflected; conversely, if a user browses cold technical resources, a general probability may determine the area of interest to the user.
The user semantic description module models the user in the form of tags. Firstly, the top N keywords with the highest scores are extracted by a TextRank algorithm when each technical resource is uploaded and are used as tags of the technical resource. After the user browses the technical resource, the same label as the technical resource label in the user label is added with a certain weight (if the label is not in the user label, the label is added).
The method comprises the following specific steps:
and extracting the technical resource key words by adopting a TextRank algorithm because the technical resources are stored and presented in a text form. Specifically, the top N keywords with the highest score are obtained as the tags of the technical resource, and the score of each keyword is normalized as the weight of the corresponding tag.
Assume that user u has a label set of LuThe label set of the technical resource i is LiThe browsing volume of the technical resource i is SiThe weight of the label a in the technical resource i is wiaThen the weight of tag a in user u is increased by a value Δ wuaIs composed of
Figure BDA0003019923010000031
If the user does not browse the technical resources related to a certain label within a period of time, the weight of the label is attenuated, so as to reflect the recent interest of the user. Assuming that a user has t time units and does not browse technical resources of a certain label, the weight before attenuation of the label is w, the weight attenuation factor is alpha, and the weight after attenuation is w
Figure BDA0003019923010000032
Because some users like a large amount of browsing technical resources, and the number of user labels is large, the first K labels with the maximum user label weight form a semantic description vector of the user.
The personalized technical resource recommendation module takes the user semantic description vector as input, calculates M technical resources most similar to the user semantic description vector through cosine similarity, and adds the M technical resources into a technical resource recommendation list of the user.
The input of the personalized technical resource recommendation module is a semantic description vector V of a user uuLet the weight of tag a in the vector be wuaThe weight of the label a in the technical resource i is wiaThen the preference degree p (u, i) of the user u to the technical resource i is
Figure BDA0003019923010000033
And after the preference degrees of the user to all the technical resources are obtained through calculation, recommending the top M technical resources with the highest preference degrees to the user.
The enterprise similarity calculation module comprises two similarity calculation methods which are respectively used for calculating the enterprise similarity by using enterprise registration information; and calculating the enterprise similarity by using the contact ratio of the enterprise user group. When the first method is used for recommending the system to be started in a cold state, the newly added enterprises cannot calculate the similarity of the enterprises by using the second method because the newly added enterprises do not have client groups.
The first method inputs enterprise information vectors obtained by a data preprocessing module, and the similarity of two enterprises is obtained through cosine similarity calculation. Assume that the information vector of enterprise e1 is Ve1The information vector of enterprise e2 is Ve2Then degree of similarity
Figure BDA0003019923010000041
And calculating the user group contact ratio between the two enterprises as the enterprise similarity, wherein the user group contact ratio can be calculated by using the Jaccard similarity. Specifically, the user group set input into the enterprise, that is, the users who have used the service provided by the enterprise, obtains the similarity of the two enterprises through the Jaccard similarity calculation. Suppose the user population of enterprise e1 is Ue1The user group of the enterprise e2 is Ue2Then degree of similarity
Figure BDA0003019923010000042
The personalized service recommendation module realizes the functions based on the similar enterprise calculation module. And according to the enterprise to which the service historically purchased by the user belongs, obtaining N enterprises closest to the enterprise through an enterprise similarity calculation module, and recommending hot services of the N enterprises to the user.
As shown in fig. 2:
the technical resource and service potential combination mining module firstly searches a cluster of user browsing technical resources near the time point according to the time of purchasing enterprise service by the user, and stores the service ID and the technical resource ID in the cluster into a set.
After a large number of sets consisting of the service IDs and the technical resource IDs are obtained, a frequent item set is obtained through calculation of an association rule algorithm FPgrowth (existing algorithm), the sets indicate technical resources which can be browsed by a plurality of users before and after a certain service is purchased, and this indicates that the service and the technical resources have strong correlation, namely, the service and the technical resources can be combined.
On the basis of obtaining a plurality of frequent itemses, if a user browses technical resources in the frequent itemsets, recommending corresponding enterprise services to the user; and if the user purchases the enterprise service, recommending the corresponding technical resource to the user.

Claims (6)

1.一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，包括数据预处理模块、用户语义描述模块、个性化技术资源推荐模块、企业相似度计算模块、个性化服务推荐模块、技术资源及服务潜在组合挖掘模块；1. a technical resource and service recommendation system based on a hybrid recommendation algorithm, characterized in that it comprises a data preprocessing module, a user semantic description module, a personalized technical resource recommendation module, an enterprise similarity calculation module, a personalized service recommendation module, Potential combination mining module of technical resources and services; 所述数据预处理模块以用户浏览技术资源记录、用户购买企业服务记录、企业注册信息为输入，其中：The data preprocessing module takes user browsing technical resource records, user purchasing enterprise service records, and enterprise registration information as input, wherein: 对用户浏览技术资源记录预处理得到向量化的记录数据和聚类簇，前者用于用户语义描述模块，后者用于技术资源及服务潜在组合挖掘模块；The vectorized record data and clusters are obtained by preprocessing the user browsing technical resource records. The former is used for the user semantic description module, and the latter is used for the potential combination mining module of technical resources and services; 对用户购买企业服务记录进行预处理得到聚类簇用于技术资源及服务潜在组合挖掘模块；Preprocessing the user's purchase of enterprise service records to obtain clusters for the mining module of potential combinations of technical resources and services; 对企业注册信息进行预处理用于企业相似度计算模块；Preprocess enterprise registration information for enterprise similarity calculation module; 所述用户语义描述模块以用户浏览技术资源记录为输入，通过用户浏览的技术资源对用户可能感兴趣的领域信息进行建模，输出用户语义描述向量用于个性化技术资源推荐模块；The user semantic description module takes the user browsing technical resource records as input, models the domain information that the user may be interested in through the technical resources browsed by the user, and outputs the user semantic description vector for the personalized technical resource recommendation module; 所述个性化技术资源推荐模块以用户语义描述向量作为输入，输出每个用户个性化的技术资源推荐列表；The personalized technical resource recommendation module takes the user semantic description vector as input, and outputs a personalized technical resource recommendation list for each user; 所述企业相似度计算模块包括两种相似度计算方法；第一种是基于内容的企业相似度计算，其输入为数据预处理模块得到的企业信息向量，在推荐系统冷启动时使用；第二种是数据驱动的企业相似度计算，其主要在企业有一定客户群体之后使用；两种算法都会得到企业之间的相似度用于个性化服务推荐模块；The enterprise similarity calculation module includes two similarity calculation methods; the first is a content-based enterprise similarity calculation, the input of which is the enterprise information vector obtained by the data preprocessing module, and is used when the recommendation system is cold-started; One is the data-driven enterprise similarity calculation, which is mainly used after the enterprise has a certain customer group; both algorithms will obtain the similarity between enterprises for the personalized service recommendation module; 所述个性化服务推荐模块以企业相似度为输入，输出每个用户个性化的的企业服务推荐列表；The personalized service recommendation module takes the enterprise similarity as an input, and outputs a personalized enterprise service recommendation list for each user; 所述技术资源及服务潜在组合挖掘模块的输入为数据预处理模块得到的用户浏览技术资源记录数据聚类簇和用户购买企业服务记录；通过关联规则算法挖掘技术资源与企业服务之间可能存在的某种组合关系，并在此基础上向用户进行推荐，输出每个用户个性化的技术资源推荐列表和企业服务推荐列表。The input of the technical resource and service potential combination mining module is the user browsing technical resource record data clusters obtained by the data preprocessing module and the user purchasing enterprise service records; the association rule algorithm is used to mine the possible existence between technical resources and enterprise services. A certain combination relationship, and recommends to users on this basis, and outputs each user's personalized technical resource recommendation list and enterprise service recommendation list. 2.根据权利要求1所述的一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，对用户浏览技术资源记录、用户购买企业服务记录等数据的记录时间进行基于密度的聚类，对企业注册信息进行数值化、编码等操作。2. a kind of technology resource and service recommendation system based on hybrid recommendation algorithm according to claim 1, is characterized in that, carries out density-based clustering to the record time of data such as user browsing technology resource record, user purchasing enterprise service record etc. , to digitize and encode the enterprise registration information. 3.根据权利要求1所述的一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，标签向量通过TextRank算法自动提取得到。3. A technology resource and service recommendation system based on a hybrid recommendation algorithm according to claim 1, wherein the label vector is automatically extracted by the TextRank algorithm. 4.根据权利要求3所述的一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，使用标签向量作为技术资源和用户的语义描述。4 . The technology resource and service recommendation system based on a hybrid recommendation algorithm according to claim 3 , wherein the tag vector is used as the semantic description of the technology resource and the user. 5 . 5.根据权利要求1所述的一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，冷启动阶段，使用企业注册信息计算企业相似度；积累了客户群里后，使用企业用户群体重合度计算企业相似度。5. A technology resource and service recommendation system based on a hybrid recommendation algorithm according to claim 1, characterized in that, in the cold start stage, the enterprise registration information is used to calculate the enterprise similarity; after accumulating the customer groups, the enterprise users are used Group coincidence calculates enterprise similarity. 6.根据权利要求1所述的一种基于混合推荐算法的技术资源及服务推荐系统，其特征在于，使用关联规则算法FPGrowth挖掘技术资源及服务的潜在组合，并在此基础上向用户推荐技术资源及服务。6. A kind of technology resource and service recommendation system based on hybrid recommendation algorithm according to claim 1, it is characterized in that, use association rule algorithm FPGrowth to mine the potential combination of technology resource and service, and recommend technology to user on this basis resources and services.
CN202110399613.7A 2021-04-14 2021-04-14 A Technology Resource and Service Recommendation System Based on Hybrid Recommendation Algorithm Active CN113076478B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110399613.7A CN113076478B (en) 2021-04-14 2021-04-14 A Technology Resource and Service Recommendation System Based on Hybrid Recommendation Algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110399613.7A CN113076478B (en) 2021-04-14 2021-04-14 A Technology Resource and Service Recommendation System Based on Hybrid Recommendation Algorithm

Publications (2)

Publication Number Publication Date
CN113076478A true CN113076478A (en) 2021-07-06
CN113076478B CN113076478B (en) 2022-06-07

Family

ID=76618372

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110399613.7A Active CN113076478B (en) 2021-04-14 2021-04-14 A Technology Resource and Service Recommendation System Based on Hybrid Recommendation Algorithm

Country Status (1)

Country Link
CN (1) CN113076478B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017128433A1 (en) * 2016-01-31 2017-08-03 胡明祥 Information pushing method during recommendation update, and pushing system
CN110188208A (en) * 2019-06-04 2019-08-30 河海大学 A method and system for querying and recommending information resources based on knowledge graphs
CN112015987A (en) * 2020-08-28 2020-12-01 青岛格兰德信用管理咨询有限公司 Potential customer recommendation system and method based on enterprise tags
CN112100512A (en) * 2020-04-10 2020-12-18 南京邮电大学 A collaborative filtering recommendation method based on user clustering and item association analysis

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017128433A1 (en) * 2016-01-31 2017-08-03 胡明祥 Information pushing method during recommendation update, and pushing system
CN110188208A (en) * 2019-06-04 2019-08-30 河海大学 A method and system for querying and recommending information resources based on knowledge graphs
CN112100512A (en) * 2020-04-10 2020-12-18 南京邮电大学 A collaborative filtering recommendation method based on user clustering and item association analysis
CN112015987A (en) * 2020-08-28 2020-12-01 青岛格兰德信用管理咨询有限公司 Potential customer recommendation system and method based on enterprise tags

Also Published As

Publication number Publication date
CN113076478B (en) 2022-06-07

Similar Documents

Publication Publication Date Title
CN112836130B (en) Context-aware recommendation system and method based on federated learning
CN112269805B (en) Data processing method, device, equipment and medium
Marinho et al. Social tagging recommender systems
CN111274330B (en) Target object determination method and device, computer equipment and storage medium
US10795895B1 (en) Business data lake search engine
CN111259263A (en) Article recommendation method and device, computer equipment and storage medium
Dhingra et al. Spam analysis of big reviews dataset using Fuzzy Ranking Evaluation Algorithm and Hadoop
US10824806B2 (en) Counterintuitive recommendations based upon temporary conditions
CN107563833A (en) A kind of personalized recommendation method and system based on block chain integration service platform
CN113722611A (en) Method, device and equipment for recommending government affair service and computer readable storage medium
CN104965883B (en) A kind of personalized travel information screening technique of matching user characteristics
CN113052653A (en) Financial product content recommendation method and system and computer readable storage medium
US10474670B1 (en) Category predictions with browse node probabilities
Cao et al. Web API recommendation via combining graph attention representation and deep factorization machines quality prediction
US20190019217A1 (en) Group formation and recommendations based on trigger events
Zhao et al. Personalized recommendation by exploring social users’ behaviors
CN109829593B (en) Credit determining method and device for target object, storage medium and electronic device
Lei et al. Personalized item recommendation algorithm for outdoor sports
CN115221954B (en) User portrait method, device, electronic equipment and storage medium
CN113076478B (en) A Technology Resource and Service Recommendation System Based on Hybrid Recommendation Algorithm
Yang An active recommendation approach to improve book-acquisition process
Bonomo et al. Customer recommendation based on profile matching and customized campaigns in on-line social networks
Kang Personalized recommendation system of smart library based on deep learning
Li Research on e-business requirement information resource extraction method in network big data
Vaganov et al. A comparative study of social data similarity measures related to financial behavior

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