WO2022222491A1 - Procédé et système de recommandation enrichie en connaissances - Google Patents

Procédé et système de recommandation enrichie en connaissances Download PDF

Info

Publication number
WO2022222491A1
WO2022222491A1 PCT/CN2021/137585 CN2021137585W WO2022222491A1 WO 2022222491 A1 WO2022222491 A1 WO 2022222491A1 CN 2021137585 W CN2021137585 W CN 2021137585W WO 2022222491 A1 WO2022222491 A1 WO 2022222491A1
Authority
WO
WIPO (PCT)
Prior art keywords
representation
semantic
knowledge
user
item
Prior art date
Application number
PCT/CN2021/137585
Other languages
English (en)
Chinese (zh)
Inventor
吕子钰
乔宇
Original Assignee
中国科学院深圳先进技术研究院
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 中国科学院深圳先进技术研究院 filed Critical 中国科学院深圳先进技术研究院
Publication of WO2022222491A1 publication Critical patent/WO2022222491A1/fr

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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Definitions

  • the present invention relates to the field of computer technology, and more particularly, to a knowledge-enhancing recommendation method and system.
  • the recommendation system plays a vital role and is an effective tool to solve the information overload.
  • the recommendation system uses the historical data of users and items and various types of other knowledge, and adopts different recommendation methods to recommend new items to users according to the user's environment and needs.
  • the recommendation system has a very wide range of application value, playing a very important role in e-commerce platforms such as Amazon, Taobao, etc., and providing users with indispensable recommendation services on multimedia platforms such as Nextfix, Youtube, and Tencent Video.
  • Traditional recommendation methods usually include content-based recommendation methods and collaborative filtering methods.
  • the content-based recommendation method relies on possible content information, such as the user's personal information or the description content information of the recommended items, etc., to estimate the user's behavior pattern and make relevant recommendations.
  • the collaborative filtering method approximates the interests of target users by analyzing the behavior patterns of the public and using the preferences of groups with similar interests or the same experience.
  • Collaborative filtering includes storage-based methods and model-based methods.
  • the storage-based method uses the user's historical behavior data to calculate the similarity or correlation for recommendation, such as the user-based collaborative filtering algorithm.
  • the model-based method is to design a machine learning model to mine behavior patterns in user behavior for recommendation.
  • the classic methods include Matrix factorization, Probabilistic Matrix Factorization and so on.
  • a research result uses a deep learning model autoencoder to learn the feature representation of users or items, and then uses collaborative filtering to learn and mine user behavior patterns.
  • a research result proposes a deep neural network structure that can integrate collaborative filtering and semi-supervised learning. Collaborative filtering and semi-supervised learning are jointly trained to achieve recommendation prediction.
  • Another research result proposes a neural collaborative recommendation algorithm to predict user behavior interaction scores for the Top-N (top N) item recommendation task.
  • the purpose of the present invention is to overcome the above-mentioned defects of the prior art, and to provide a knowledge enhancement recommendation method and system, which utilizes an external knowledge base to perform knowledge enhancement in three aspects of users, items, and user-item interaction, and learns the knowledge enhancement representation of user behavior. , to alleviate the impact of data sparsity and solve the cold start problem.
  • a knowledge-enhancing recommendation method includes the following steps:
  • Step S1 For the review document, extract the global-level semantic vector representation and the sentence-level implicit vector representation to obtain a hierarchical semantic representation
  • Step S2 using the knowledge base to retrieve the semantic knowledge associated with the review document to form an associated semantic knowledge representation
  • Step S3 fusing the hierarchical semantic knowledge and the associated semantic knowledge representation to obtain the enhanced representation of the semantic knowledge of the user, the enhanced representation of the semantic knowledge of the item, and the enhanced representation of the semantic knowledge of the user-item interaction;
  • Step S4 Based on the neural collaborative filtering framework, the obtained semantic knowledge enhanced representation of the user, the semantic knowledge enhanced representation of the item, and the semantic knowledge enhanced representation of the user-item interaction are fused, and the interaction score is predicted by using the hierarchical user-item interaction mechanism. to generate item recommendation results.
  • a knowledge-enhanced recommendation system includes:
  • Hierarchical semantic extraction unit It is used to extract global-level semantic vector representation and sentence-level implicit vector representation for review documents, and obtain hierarchical semantic representation;
  • Associative semantic extraction unit used to retrieve the semantic knowledge associated with the review document by using the knowledge base, and form the associated semantic knowledge representation
  • Knowledge enhancement representation unit used to fuse the hierarchical semantic knowledge and the associated semantic knowledge representation to obtain the enhanced semantic knowledge representation of the user, the enhanced representation of the semantic knowledge of the item, and the enhanced representation of the semantic knowledge of the user-item interaction;
  • Recommendation prediction unit Based on the neural collaborative filtering framework, it is used to integrate the obtained semantic knowledge enhanced representation of the user, the semantic knowledge enhanced representation of the item, and the semantic knowledge enhanced representation of the user-item interaction, and use the hierarchical user-item interaction mechanism to predict Interactive scoring to generate item recommendation results.
  • the present invention has the advantages that, aiming at the severe data sparse problem in the recommendation system, external knowledge is used to enhance semantic knowledge in three aspects of user, item and user-item interaction, and hierarchical knowledge is designed to enhance user behavior.
  • Representation learning method obtain high-quality feature representation, accurately describe user behavior preferences, solve the problem of data sparse and cold start; based on neural collaborative filtering framework, integrate knowledge-enhanced user behavior representation, and propose a hierarchical neural network recommendation algorithm. Simulate user-item interaction to achieve recommendation prediction, provide high-quality recommendation results, and improve recommendation performance.
  • FIG. 1 is a flowchart of a recommendation method for knowledge enhancement according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of a hierarchical neural network recommendation algorithm according to an embodiment of the present invention.
  • the problem to be solved in this paper can be defined as: given user set U, item set V, user-item interaction matrix R and text comment set D, common sense knowledge base KG, for a given user u, recommend top-N items ( items that the user has not interacted with).
  • the knowledge-enhanced recommendation method provided by the present invention mainly includes two parts, namely, knowledge-enhanced user behavior representation learning and hierarchical neural network recommendation algorithm. Specifically, as shown in FIG. 1 , the method includes the following steps.
  • Step S110 learn the knowledge-enhanced user behavior representation, and obtain the knowledge-enhanced representation of the user, the knowledge-enhanced representation of the item, and the knowledge-enhanced representation of the user-item interaction.
  • the semantic knowledge enhancement process includes the following steps.
  • Step S111 extracting the semantic vector representation at the global level and the implicit vector representation at the sentence level to obtain a hierarchical semantic representation.
  • the word2vec method is used to construct the word vector, and then the global-level bidirectional long short-term memory network BiLSTM is used to extract the global-level semantic latent vector representation S g ; the sentence-level BiLSTM is used to extract the sentence-level semantics The implicit vector representation S s .
  • the semantic vector S g at the global level contains the global latent semantic information, and the latent vector representation S s at the sentence level contains a more fine-grained semantic knowledge representation with context information; the two semantic representations are fused through the concatenate operation.
  • the hierarchical semantic representation C [S g , S s ] is obtained.
  • the hierarchical semantic representation obtained in this step S111 is sometimes also referred to as a shallow semantic representation, unless otherwise indicated according to the context.
  • step S112 a deep-level semantic knowledge representation is obtained based on the knowledge base.
  • the top-N related concepts are retrieved from the common sense knowledge base KB, and the vector representation of the concepts is denoted as c.
  • a multi-head attention mechanism is used to fuse user-item hierarchical semantic representation C and enhanced semantic key-value pairs to obtain deep semantic knowledge representation.
  • the specific formula is as follows:
  • MultiHead(C,S g ,C A ) Concat(h 1 ,h 1 ,...,h h )W O
  • W O , Wi Q , Wi K , Wi A are the learning parameters of the multi-head attention mechanism. Therefore, the semantic knowledge representations of users, items, and user-item interactions are denoted as
  • Step S113 fuse the user/item shallow representation and the user/item deep semantic knowledge representation to obtain the user's enhanced semantic knowledge representation, the enhanced semantic knowledge representation of the item, and the enhanced semantic knowledge representation of the user-item interaction, respectively.
  • one-hot encoding (one-hot) is used to represent the used users and items, and the method of embedding representation is applied to map the one-hot encoding into a dense space of user shallow representation u and item shallow representation v, and then by The concatenation operation fuses the user/item shallow representation and the user/item deep semantic knowledge representation, and obtains the user's semantic knowledge enhancement representation respectively and Semantic Knowledge Augmented Representations of Items Semantic Knowledge Augmented Representations of User-Item Interactions, labeled as
  • this step S110 semantic analysis is performed on the user, item, and user-item text documents to extract semantic knowledge at different levels; the associated semantic knowledge is retrieved from the common sense knowledge base to form an associated semantic knowledge representation; the hierarchical semantic knowledge is fused and associated semantic knowledge representations to learn knowledge-augmented representations of users, items, and user-item interactions for subsequent user behavior interaction prediction.
  • knowledge base information the attribute information of entities in different domains can be flexibly characterized, and the products in the recommender system and the entities in the knowledge base can be associated together, so as to obtain rich products from various fields. property information.
  • step S120 the knowledge-enhanced representation of users, items, and user-item interactions is integrated, and a hierarchical user-item interaction mechanism is used to predict interaction scores, thereby generating item recommendation results.
  • a three-layer structure of user-item interaction prediction algorithm is designed, as shown in Figure 3.
  • the first layer uses a neural network (f 1 ) to map the knowledge-augmented user representation Xu and the knowledge-augmented item representation X v to the user-item interaction representation I uv , expressed as:
  • W I is the learning weight parameter
  • b is the bias parameter
  • f 1 is the activation function, which can be selected from tanh, ReLU, etc.
  • the last layer (the third layer or prediction layer) predicts the user-item preference score If expressed as:
  • step S120 based on the neural network collaborative filtering framework, the knowledge enhancement representation of users, items, and user-item interactions is integrated, a hierarchical user-item interaction mechanism is designed to predict interaction scores, and recommendations are generated based on the user-item interaction prediction scores. result.
  • the method proposed by the present invention is an end-to-end method. Therefore, the knowledge-enhanced feature representation learning and the hierarchical neural network recommendation prediction are trained and optimized by means of joint learning, which significantly reduces the training difficulty and accelerates the training process. Also reduces memory requirements.
  • the present invention also provides a knowledge-enhanced recommendation system for implementing one or more aspects of the above method.
  • the system includes: a hierarchical semantic extraction unit, which is used for extracting a global level for review documents The semantic vector representation and the sentence-level implicit vector representation of , obtain hierarchical semantic representation; the associated semantic extraction unit is used to use the knowledge base to retrieve the semantic knowledge associated with the review document, and form the associated semantic knowledge representation; knowledge enhancement representation a unit, which is used to fuse the hierarchical semantic knowledge and the associated semantic knowledge representation to obtain the enhanced representation of the semantic knowledge of the user, the enhanced representation of the semantic knowledge of the item, and the enhanced representation of the semantic knowledge of the user-item interaction; the recommendation prediction unit, which Based on the neural collaborative filtering framework, it is used to fuse the obtained semantic knowledge-enhanced representation of users, the semantic knowledge-enhanced representation of items, and the semantic-knowledge-enhanced representation of user-item interaction, and use the hierarchical user-item interaction mechanism to
  • the present invention proposes a technical solution for solving the problem of sparse user behavior data by using knowledge enhancement, using an external semantic knowledge base to enhance knowledge at three levels of user, item, and user-item interaction, and learn the knowledge-enhanced representation of user behavior. , effectively reducing the impact of data sparseness and solving the cold start problem.
  • a hierarchical neural network recommendation algorithm is innovatively proposed, which mines user behavior patterns from different levels, simulates user behavior decision-making, achieves high-quality user-item interaction prediction, and improves recommendation performance.
  • bidirectional GRU Gate Recurrent Unit
  • unidirectional LSTM unidirectional GRU
  • unidirectional GRU unidirectional GRU
  • the purpose of preferably adopting BiLSTM is to fully learn the semantic knowledge representation of bidirectional (ie, contextual) information.
  • the present invention may be a system, method and/or computer program product.
  • the computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of the present invention.
  • a computer-readable storage medium may be a tangible device that can hold and store instructions for use by the instruction execution device.
  • the computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing.
  • Non-exhaustive list of computer readable storage media include: portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM) or flash memory), static random access memory (SRAM), portable compact disk read only memory (CD-ROM), digital versatile disk (DVD), memory sticks, floppy disks, mechanically coded devices, such as printers with instructions stored thereon Hole cards or raised structures in grooves, and any suitable combination of the above.
  • RAM random access memory
  • ROM read only memory
  • EPROM erasable programmable read only memory
  • flash memory static random access memory
  • SRAM static random access memory
  • CD-ROM compact disk read only memory
  • DVD digital versatile disk
  • memory sticks floppy disks
  • mechanically coded devices such as printers with instructions stored thereon Hole cards or raised structures in grooves, and any suitable combination of the above.
  • Computer-readable storage media are not to be construed as transient signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (eg, light pulses through fiber optic cables), or through electrical wires transmitted electrical signals.
  • the computer readable program instructions described herein may be downloaded to various computing/processing devices from a computer readable storage medium, or to an external computer or external storage device over a network such as the Internet, a local area network, a wide area network, and/or a wireless network.
  • the network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and/or edge servers.
  • a network adapter card or network interface in each computing/processing device receives computer-readable program instructions from a network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing/processing device .
  • the computer program instructions for carrying out the operations of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state setting data, or instructions in one or more programming languages.
  • Source or object code written in any combination including object-oriented programming languages, such as Smalltalk, C++, Python, etc., and conventional procedural programming languages, such as the "C" language or similar programming languages.
  • the computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server implement.
  • the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, using an Internet service provider via the Internet) connect).
  • LAN local area network
  • WAN wide area network
  • custom electronic circuits such as programmable logic circuits, field programmable gate arrays (FPGAs), or programmable logic arrays (PLAs) can be personalized by utilizing state information of computer readable program instructions.
  • Computer readable program instructions are executed to implement various aspects of the present invention.
  • These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer or other programmable data processing apparatus to produce a machine that causes the instructions when executed by the processor of the computer or other programmable data processing apparatus , resulting in means for implementing the functions/acts specified in one or more blocks of the flowchart and/or block diagrams.
  • These computer readable program instructions can also be stored in a computer readable storage medium, these instructions cause a computer, programmable data processing apparatus and/or other equipment to operate in a specific manner, so that the computer readable medium on which the instructions are stored includes An article of manufacture comprising instructions for implementing various aspects of the functions/acts specified in one or more blocks of the flowchart and/or block diagrams.
  • Computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other equipment to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other equipment to produce a computer-implemented process , thereby causing instructions executing on a computer, other programmable data processing apparatus, or other device to implement the functions/acts specified in one or more blocks of the flowcharts and/or block diagrams.
  • each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more functions for implementing the specified logical function(s) executable instructions.
  • the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations can be implemented in dedicated hardware-based systems that perform the specified functions or actions , or can be implemented in a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that implementation in hardware, implementation in software, and implementation in a combination of software and hardware are all equivalent.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

La présente invention divulgue un procédé et un système de recommandation enrichie en connaissances. Le procédé consiste : pour un document de commentaire, à extraire une représentation vectorielle sémantique de niveau global et une représentation vectorielle implicite au niveau de la phrase pour obtenir une représentation sémantique hiérarchique ; à utiliser une base de connaissances pour récupérer des connaissances sémantiques associées au document de commentaire pour former une représentation de connaissance sémantique associée ; à fusionner les connaissances sémantiques hiérarchiques et la représentation de connaissances sémantiques associées pour obtenir une représentation sémantique enrichie en connaissances pour un utilisateur, une représentation sémantique enrichie en connaissances pour un article et une représentation sémantique enrichie en connaissances de l'interaction utilisateur-article ; à fusionner la représentation sémantique enrichie en connaissances obtenue pour l'utilisateur, la représentation sémantique enrichie en connaissances pour l'article et la représentation sémantique enrichie en connaissances pour l'interaction utilisateur-article sur la base d'un cadre de filtrage collaboratif neuronal, et à prédire un score d'interaction à l'aide d'un mécanisme d'interaction utilisateur-article hiérarchique pour générer un résultat de recommandation d'article. La présente invention simule de manière hiérarchique une interaction utilisateur-article pour réaliser une prédiction de recommandation, ce qui permet de fournir un résultat de recommandation de qualité élevée.
PCT/CN2021/137585 2021-04-22 2021-12-13 Procédé et système de recommandation enrichie en connaissances WO2022222491A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110435550.6 2021-04-22
CN202110435550.6A CN113158049B (zh) 2021-04-22 2021-04-22 一种知识增强的推荐方法和系统

Publications (1)

Publication Number Publication Date
WO2022222491A1 true WO2022222491A1 (fr) 2022-10-27

Family

ID=76869469

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/137585 WO2022222491A1 (fr) 2021-04-22 2021-12-13 Procédé et système de recommandation enrichie en connaissances

Country Status (2)

Country Link
CN (1) CN113158049B (fr)
WO (1) WO2022222491A1 (fr)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113158049B (zh) * 2021-04-22 2022-11-01 中国科学院深圳先进技术研究院 一种知识增强的推荐方法和系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100228693A1 (en) * 2009-03-06 2010-09-09 phiScape AG Method and system for generating a document representation
CN103593792A (zh) * 2013-11-13 2014-02-19 复旦大学 一种基于中文知识图谱的个性化推荐方法与系统
CN110807154A (zh) * 2019-11-08 2020-02-18 内蒙古工业大学 一种基于混合深度学习模型的推荐方法与系统
CN112231577A (zh) * 2020-11-06 2021-01-15 重庆理工大学 一种融合文本语义向量和神经协同过滤的推荐方法
CN113158049A (zh) * 2021-04-22 2021-07-23 中国科学院深圳先进技术研究院 一种知识增强的推荐方法和系统

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108921657B (zh) * 2018-06-25 2021-06-29 中国人民大学 一种基于知识增强记忆网络的序列推荐方法
CN109299396B (zh) * 2018-11-28 2020-11-06 东北师范大学 融合注意力模型的卷积神经网络协同过滤推荐方法及系统
CN110084670B (zh) * 2019-04-15 2022-03-25 东北大学 一种基于lda-mlp的货架商品组合推荐方法
CN111523029B (zh) * 2020-04-20 2022-03-25 浙江大学 一种基于知识图谱表示学习的个性化推荐方法
CN112100485A (zh) * 2020-08-20 2020-12-18 齐鲁工业大学 一种基于评论的评分预测物品推荐方法及系统
CN112115358B (zh) * 2020-09-14 2024-04-16 中国船舶重工集团公司第七0九研究所 一种利用知识图谱中多跳路径特征的个性化推荐方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100228693A1 (en) * 2009-03-06 2010-09-09 phiScape AG Method and system for generating a document representation
CN103593792A (zh) * 2013-11-13 2014-02-19 复旦大学 一种基于中文知识图谱的个性化推荐方法与系统
CN110807154A (zh) * 2019-11-08 2020-02-18 内蒙古工业大学 一种基于混合深度学习模型的推荐方法与系统
CN112231577A (zh) * 2020-11-06 2021-01-15 重庆理工大学 一种融合文本语义向量和神经协同过滤的推荐方法
CN113158049A (zh) * 2021-04-22 2021-07-23 中国科学院深圳先进技术研究院 一种知识增强的推荐方法和系统

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
HONGWEI WANG ; FUZHENG ZHANG ; MIAO ZHAO ; WENJIE LI ; XING XIE ; MINYI GUO: "Multi-Task Feature Learning for Knowledge Graph Enhanced Recommendation", THE WORLD WIDE WEB CONFERENCE, 13 May 2019 (2019-05-13), pages 2000 - 2010, XP058434079, ISBN: 978-1-4503-6674-8, DOI: 10.1145/3308558.3313411 *
WU XIYU;CHEN QIMAI;LIU HAI;HE CHAOBO: "Collaborative Filtering Recommendation Algorithm Based on Representation Learning of Knowledge Graph", COMPUTER ENGINEERING, vol. 44, no. 2, 21 April 2017 (2017-04-21), pages 226 - 232+263, XP055977837, ISSN: 1000-3428, DOI: 10.3969/j.issn.1000-3428.2018.02.039 *

Also Published As

Publication number Publication date
CN113158049B (zh) 2022-11-01
CN113158049A (zh) 2021-07-23

Similar Documents

Publication Publication Date Title
Torfi et al. Natural language processing advancements by deep learning: A survey
US11182562B2 (en) Deep embedding for natural language content based on semantic dependencies
Young et al. Recent trends in deep learning based natural language processing
US10922488B1 (en) Computing numeric representations of words in a high-dimensional space
US20200380023A1 (en) Classifying data objects
Wang et al. Combining Knowledge with Deep Convolutional Neural Networks for Short Text Classification.
Britz Understanding convolutional neural networks for NLP
WO2022222037A1 (fr) Procédé de recommandation interprétable basé sur l'inférence de réseau neuronal de graphe
US20200193245A1 (en) Aligning symbols and objects using co-attention for understanding visual content
US20240185602A1 (en) Cross-Modal Processing For Vision And Language
CN111783903B (zh) 文本处理方法、文本模型的处理方法及装置、计算机设备
Suman et al. Why pay more? A simple and efficient named entity recognition system for tweets
Hui et al. Few-shot relation classification by context attention-based prototypical networks with BERT
Lin et al. Reliability-aware dynamic feature composition for name tagging
KR20210034679A (ko) 엔티티-속성 관계 식별
He et al. Dual long short-term memory networks for sub-character representation learning
CN114416995A (zh) 信息推荐方法、装置及设备
WO2022222491A1 (fr) Procédé et système de recommandation enrichie en connaissances
JP2023002690A (ja) セマンティックス認識方法、装置、電子機器及び記憶媒体
Zhang et al. SKG-Learning: A deep learning model for sentiment knowledge graph construction in social networks
CN113961666B (zh) 关键词识别方法、装置、设备、介质及计算机程序产品
Wei et al. Sentiment classification of tourism reviews based on visual and textual multifeature fusion
Kroon et al. Advancing Automated Content Analysis for a New Era of Media Effects Research: The Key Role of Transfer Learning
Liu et al. Sentence representation
CN115858732A (zh) 实体链接方法及设备

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: 21937713

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: 21937713

Country of ref document: EP

Kind code of ref document: A1