Atnaujinkite slapukų nuostatas

Social Web Artifacts for Boosting Recommenders: Theory and Implementation 2013 ed. [Kietas viršelis]

4.00/5 (2 ratings by Goodreads)
  • Formatas: Hardback, 187 pages, aukštis x plotis: 235x155 mm, weight: 501 g, XIX, 187 p., 1 Hardback
  • Serija: Studies in Computational Intelligence 487
  • Išleidimo metai: 31-May-2013
  • Leidėjas: Springer International Publishing AG
  • ISBN-10: 331900526X
  • ISBN-13: 9783319005263
Kitos knygos pagal šią temą:
Social Web Artifacts for Boosting Recommenders: Theory and Implementation 2013 ed.Didesnis paveikslėlis
  • Formatas: Hardback, 187 pages, aukštis x plotis: 235x155 mm, weight: 501 g, XIX, 187 p., 1 Hardback
  • Serija: Studies in Computational Intelligence 487
  • Išleidimo metai: 31-May-2013
  • Leidėjas: Springer International Publishing AG
  • ISBN-10: 331900526X
  • ISBN-13: 9783319005263
Kitos knygos pagal šią temą:
Pastovi nuoroda: https://www.kriso.lt/db/9783319005263.html
Raktažodžiai:
Recommender systems, software programs that learn from human behavior and make predictions of what products we are expected to appreciate and purchase, have become an integral part of our everyday life. They proliferate across electronic commerce around the globe and exist for virtually all sorts of consumable goods, such as books, movies, music, or clothes.At the same time, a new evolution on the Web has started to take shape, commonly known as the “Web 2.0” or the “Social Web”: Consumer-generated media has become rife, social networks have emerged and are pulling significant shares of Web traffic. In line with these developments, novel information and knowledge artifacts have become readily available on the Web, created by the collective effort of millions of people.This textbook presents approaches to exploit the new Social Web fountain of knowledge, zeroing in first and foremost on two of those information artifacts, namely classification taxonomies and trust networks. These two are used to improve the performance of product-focused recommender systems: While classification taxonomies are appropriate means to fight the sparsity problem prevalent in many productive recommender systems, interpersonal trust ties – when used as proxies for interest similarity – are able to mitigate the recommenders' scalability problem.

This book presents approaches for exploiting the rapidly expanding fountain of Social Web knowledge by means of classification taxonomies and trust networks, which are used to improve the performance of product-focused recommender systems.
Part I Laying Foundations
1 Introduction
3(8)
1.1 Motivation
3(3)
1.1.1 Collaborative Filtering Systems
3(1)
1.1.2 The Participation Age on the Web
4(2)
1.2 Organization and Content
6(1)
1.3 Contributions
6(5)
1.3.1 Publications
9(2)
2 On Recommender Systems
11(12)
2.1 Introduction
11(1)
2.2 Collecting Preference Information
11(1)
2.3 Recommender System Types and Techniques
12(5)
2.3.1 Content-Based Techniques
12(1)
2.3.2 Collaborative Filtering
13(1)
2.3.2.1 User-Based Collaborative Filtering
14(2)
2.3.2.2 Item-Based Collaborative Filtering
16(1)
2.3.3 Hybrid Recommender Systems
16(1)
2.4 Evaluating Recommender Systems
17(6)
2.4.1 Accuracy Metrics
17(1)
2.4.1.1 Predictive Accuracy Metrics
18(1)
2.4.1.2 Decision-Support Metrics
18(2)
2.4.2 Beyond Accuracy
20(1)
2.4.2.1 Coverage
20(1)
2.4.2.2 Novelty and Serendipity
20(3)
Part II Use of Taxonomic Knowledge
3 Taxonomy-Driven Filtering
23(24)
3.1 Introduction
23(1)
3.2 Related Work
24(1)
3.3 Approach Outline
24(8)
3.3.1 Information Model
25(1)
3.3.2 Taxonomy-Driven Profile Generation
25(2)
3.3.3 Neighborhood Formation
27(1)
3.3.3.1 Measuring Proximity
28(1)
3.3.3.2 Selecting Neighbors
29(1)
3.3.4 Recommendation Generation
29(1)
3.3.5 Topic Diversification
30(1)
3.3.5.1 Recommendation Dependency
31(1)
3.3.5.2 Topic Diversification Algorithm
32(1)
3.3.5.3 Osmotic Pressure Analogy
32(1)
3.4 Offline Experiments and Evaluation
32(6)
3.4.1 Data Acquisition
33(1)
3.4.2 Evaluation Framework
33(1)
3.4.2.1 Benchmark Systems
33(2)
3.4.2.2 Experiment Setup
35(1)
3.4.2.3 Parameterization
35(1)
3.4.2.4 Result Analysis
35(3)
3.5 Deployment and Online Study
38(2)
3.5.1 Online Study Setup
39(1)
3.5.2 Result Analysis
39(1)
3.6 Movie Data Analysis
40(4)
3.6.1 Dataset Composition
41(1)
3.6.2 Offline Experiment Framework
41(1)
3.6.2.1 Benchmark Systems
41(1)
3.6.2.2 Setup and Metrics
42(1)
3.6.2.3 Result Analysis
43(1)
3.7 Conclusion
44(3)
4 Topic Diversification Revisited
47(14)
4.1 Introduction
47(1)
4.2 Related Work
48(1)
4.3 Empirical Analysis
49(10)
4.3.1 Offline Experiments
50(1)
4.3.1.1 Experiment Setup
51(1)
4.3.1.2 Result Analysis
51(2)
4.3.2 User Survey
53(1)
4.3.2.1 Survey Outline and Setup
54(1)
4.3.2.2 Result Analysis
54(3)
4.3.2.3 Multiple Linear Regression
57(1)
4.3.3 Limitations
58(1)
4.4 Summary
59(2)
5 Taxonomies for Calculating Semantic Proximity
61(18)
5.1 Introduction
61(2)
5.1.1 On Calculating the Similarity of Word Meanings
61(1)
5.1.2 Contributions
62(1)
5.2 Related Work
63(1)
5.3 Framework
64(1)
5.3.1 Service Requirements
64(1)
5.3.2 Algorithm Outline
64(1)
5.4 Proximity Metrics
65(3)
5.4.1 Similarity between Word-Sense Pairs
65(1)
5.4.2 Multi-class Categorization Approach
66(1)
5.4.2.1 Profiling Phase
66(1)
5.4.2.2 Measuring Proximity
67(1)
5.5 Empirical Evaluation
68(9)
5.5.1 Benchmark Layout
69(1)
5.5.2 Online Survey Design
69(1)
5.5.3 Proximity Metrics
70(1)
5.5.3.1 Taxonomy-Driven Metrics
70(1)
5.5.3.2 Text-Based Approaches
71(1)
5.5.4 Experiments
71(1)
5.5.4.1 Parameter Learning
71(2)
5.5.4.2 Performance Analysis
73(3)
5.5.4.3 Conclusion
76(1)
5.6 Outlook
77(2)
6 Recommending Technology Synergies
79(20)
6.1 Introduction
79(1)
6.2 Motivation for Recommending Technology Synergies
79(2)
6.3 Scenario Description
81(1)
6.4 System Setup
81(4)
6.4.1 Wikipedia-Based Classifier
82(1)
6.4.2 ODP-Based Classifier
83(1)
6.4.3 Merging Classifiers
83(2)
6.5 System Usage
85(3)
6.5.1 Pre-processing
85(1)
6.5.2 Pivot Tabulation
86(1)
6.5.3 Hypothesis Validation
87(1)
6.6 Synergy Classifier Evaluation
88(6)
6.6.1 Experiment Setup
88(3)
6.6.2 Result Analysis
91(1)
6.6.2.1 Intra-group Correlation
91(1)
6.6.2.2 Inter-group Correlation
92(1)
6.6.2.3 Classifier Benchmarking
92(1)
6.6.3 Conclusion
93(1)
6.7 Related Work
94(1)
6.8 Discussion
94(5)
Part III Social Ties and Trust
7 Trust Propagation Models
99(34)
7.1 Introduction
99(1)
7.2 Contributions
99(2)
7.2.1 On Trust and Trust Propagation
100(1)
7.3 Computational Trust in Social Networks
101(6)
7.3.1 Classification of Trust Metrics
101(3)
7.3.2 Trust and Decentralization
104(1)
7.3.2.1 Trust Model
105(1)
7.3.2.2 Trust Metrics for Decentralized Networks
105(2)
7.4 Local Group Trust Metrics
107(17)
7.4.1 Outline of Advogato Maxflow
107(1)
7.4.1.1 Trust Computation Steps
107(3)
7.4.1.2 Attack-Resistance Properties
110(1)
7.4.2 The Appleseed Trust Metric
111(1)
7.4.2.1 Searches in Contextual Network Graphs
111(1)
7.4.2.2 Trust Propagation
111(1)
7.4.2.3 Spreading Factor
112(2)
7.4.2.4 Rank Normalization
114(1)
7.4.2.5 Backward Trust Propagation
115(1)
7.4.2.6 Nonlinear Trust Normalization
115(1)
7.4.2.7 Algorithm Outline
116(1)
7.4.2.8 Parameterization and Experiments
117(5)
7.4.2.9 Implementation and Extensions
122(1)
7.4.3 Comparison of Advogato and Appleseed
123(1)
7.5 Distrust
124(7)
7.5.1 Semantics of Distrust
125(1)
7.5.1.1 Distrust as Negated Trust
125(1)
7.5.1.2 Propagation of Distrust
126(1)
7.5.2 Incorporating Distrust into Appleseed
126(1)
7.5.2.1 Normalization and Distrust
127(1)
7.5.2.2 Distrust Allocation and Propagation
128(1)
7.5.2.3 Convergence
129(2)
7.6 Discussion and Outlook
131(2)
8 Interpersonal Trust and Similarity
133(22)
8.1 Introduction
133(1)
8.2 Trust Models in Recommender Systems
134(1)
8.3 Evidence from Social Psychology
135(3)
8.3.1 On Interpersonal Attraction and Similarity
135(1)
8.3.1.1 The Bogus Stranger Technique
136(1)
8.3.1.2 Analysis of Similarity-Attraction Associations
136(1)
8.3.1.3 Limitations
137(1)
8.3.2 Conclusion
137(1)
8.4 Trust-Similarity Correlation Analysis
138(7)
8.4.1 Model and Data Acquisition
138(1)
8.4.1.1 Information Model
138(1)
8.4.1.2 Data Acquisition
139(1)
8.4.2 Experiment Setup and Analysis
139(1)
8.4.2.1 Upper Bound Analysis
139(1)
8.4.2.2 Lower Bound Analysis
140(3)
8.4.3 Statistical Significance
143(1)
8.4.4 Conclusion
144(1)
8.5 Trust and Similarity in FilmTrust
145(3)
8.5.1 FilmTrust Introduction
145(2)
8.5.2 Profile Similarity Computation
147(1)
8.5.3 Statistical Significance
147(1)
8.5.4 Conclusion
148(1)
8.6 Exploiting Correlations between Trust and Similarity
148(3)
8.7 Discussion and Outlook
151(4)
Part IV Amalgamating Taxonomies and Trust
9 Decentralized Recommender Systems
155(18)
9.1 Introduction
155(1)
9.2 On Decentralized Recommender Systems
155(2)
9.2.1 Decentralized Recommender Setup
156(1)
9.2.2 Research Issues
156(1)
9.3 Principal Design Considerations
157(2)
9.4 Related Work
159(1)
9.5 Framework Components
160(4)
9.5.1 Trust-Based Neighborhood Formation
160(1)
9.5.1.1 Network Connectivity
160(2)
9.5.1.2 Trust Propagation Models
162(1)
9.5.2 Measuring User Similarity and Product-User Relevance
163(1)
9.5.3 Recommendation Generation
163(1)
9.6 Offline Experiments and Evaluation
164(6)
9.6.1 Dataset Acquisition
164(1)
9.6.2 Evaluation Framework
165(1)
9.6.2.1 Experiment Setup
166(1)
9.6.2.2 Parameterization
166(1)
9.6.3 Experiments
166(1)
9.6.3.1 Neighborhood Formation Impact
166(2)
9.6.3.2 Neighborhood Size Sensitivity
168(1)
9.6.3.3 Neighborhood Overlap Analysis
169(1)
9.7 Conclusion and Outlook
170(3)
10 Conclusion
173(4)
10.1 Summary
173(1)
10.2 Discussion and Outlook
174(3)
References 177