一種使用網(wǎng)絡(luò)嵌入方法的知識驅(qū)動的紙質(zhì)推薦方法
發(fā)布時間:2022-08-10 20:55
日常生活中,我們經(jīng)常在互聯(lián)網(wǎng)上搜索各種各樣的東西,而且有很多搜索引擎能夠?yàn)槲覀兯阉鞯较嚓P(guān)的結(jié)果。隨著技術(shù)的飛速發(fā)展,互聯(lián)網(wǎng)已變成了人們獲取信息的主要來源。此外,Web2.0時代的到來使得用戶和網(wǎng)站之間的交互增加了。依據(jù)用戶興趣為用戶提供信息,變得具有挑戰(zhàn)性。然而由于版權(quán)的限制,現(xiàn)有的大多數(shù)研究都面臨著缺乏候選推薦文章內(nèi)容的問題。這些文章的內(nèi)容并不都是可以免費(fèi)獲得的,這導(dǎo)致了推薦結(jié)果不充分。此外,很多研究是基于推薦用戶的個人資料的。因此,他們的推薦需要系統(tǒng)中有大量的注冊用戶。近年來,研究工作證明知識圖在產(chǎn)生高質(zhì)量推薦結(jié)果、減輕稀疏性和冷啟動問題方面取得了更好的效果。網(wǎng)絡(luò)嵌入技術(shù)嘗試從網(wǎng)絡(luò)結(jié)構(gòu)中自動地學(xué)習(xí)高質(zhì)量特征向量,使得基于向量的節(jié)點(diǎn)相關(guān)性度量成為可能。保持網(wǎng)絡(luò)嵌入技術(shù)的優(yōu)勢,本文提出了知識驅(qū)動的論文推薦方法,利用異構(gòu)網(wǎng)絡(luò)嵌入模型來生成推薦結(jié)果。本文的創(chuàng)新性在于利用網(wǎng)絡(luò)嵌入方法的性能,即matapath2vec,此方法在學(xué)習(xí)知識網(wǎng)絡(luò)并找到滿足用戶需求的論文以及識別并整合論文中的潛在關(guān)系方面,可以發(fā)揮重要作用,這可以幫助改善推薦的結(jié)果。與現(xiàn)有方法不同,本文所提出的方法具有學(xué)習(xí)網(wǎng)絡(luò)中節(jié)點(diǎn)(...
【文章頁數(shù)】:64 頁
【學(xué)位級別】:碩士
【文章目錄】:
摘要
Abstract
Abbreviations& Acronyms
Chapter1 Introduction
1.1 Background
1.2 Research background and research significance
1.3 Research content and contribution
1.4 Organization of the thesis
Chapter2 Related Work
2.1 Traditional recommendation Algorithm
2.2 A content-based Paper Recommender System
2.3 Knowledge Graph Embeddings Based Recommendation Using Node2vec
2.4 Recommendation model based on knowledge network representation
2.5 Summary
Chapter3 The KN-HER Model
3.1 Problem formulation
3.2 Problem Definition
3.3 Architecture and working of the proposed method
3.4 Model Overview
3.4.1 Construction of Heterogeneous Papers Network
3.4.2 Heterogeneous Papers Network Embedding
3.4.3 KN-HER Model based Citation Recommendation Approach
3.5 Cold-start and Sparsity scenario in paper recommendation
3.6 Summary
Chapter4 Performance Evaluation and Results
4.1 Experimental settings
4.1.1 Implementation plateform
4.1.2 Implemented Dataset
4.2 Evaluation Metrics
4.3 Evaluation Results
4.3.1 Analysis on the basis of Precision
4.3.2 Analysis on The Basis of Recall
4.3.3 Analysis on The Basis of NDCG
4.4 Case Studies
4.5 Summary
Chapter5 System Design and Implementation
5.1 System framework and design
5.1.1 System Framework
5.1.2 System Functions
5.2 Data acquisition
5.3 System implementation and function display
5.3.1 System Implementation
5.4 Summary
Chapter6 Conclusion and Future Work
6.1 Conclusion
6.2 Future work
Acknowledgement
References
本文編號:3674344
【文章頁數(shù)】:64 頁
【學(xué)位級別】:碩士
【文章目錄】:
摘要
Abstract
Abbreviations& Acronyms
Chapter1 Introduction
1.1 Background
1.2 Research background and research significance
1.3 Research content and contribution
1.4 Organization of the thesis
Chapter2 Related Work
2.1 Traditional recommendation Algorithm
2.2 A content-based Paper Recommender System
2.3 Knowledge Graph Embeddings Based Recommendation Using Node2vec
2.4 Recommendation model based on knowledge network representation
2.5 Summary
Chapter3 The KN-HER Model
3.1 Problem formulation
3.2 Problem Definition
3.3 Architecture and working of the proposed method
3.4 Model Overview
3.4.1 Construction of Heterogeneous Papers Network
3.4.2 Heterogeneous Papers Network Embedding
3.4.3 KN-HER Model based Citation Recommendation Approach
3.5 Cold-start and Sparsity scenario in paper recommendation
3.6 Summary
Chapter4 Performance Evaluation and Results
4.1 Experimental settings
4.1.1 Implementation plateform
4.1.2 Implemented Dataset
4.2 Evaluation Metrics
4.3 Evaluation Results
4.3.1 Analysis on the basis of Precision
4.3.2 Analysis on The Basis of Recall
4.3.3 Analysis on The Basis of NDCG
4.4 Case Studies
4.5 Summary
Chapter5 System Design and Implementation
5.1 System framework and design
5.1.1 System Framework
5.1.2 System Functions
5.2 Data acquisition
5.3 System implementation and function display
5.3.1 System Implementation
5.4 Summary
Chapter6 Conclusion and Future Work
6.1 Conclusion
6.2 Future work
Acknowledgement
References
本文編號:3674344
本文鏈接:http://sikaile.net/kejilunwen/shengwushengchang/3674344.html
最近更新
教材專著