Muutke küpsiste eelistusi

Relevance Ranking for Vertical Search Engines [Pehme köide]

2.50/5 (3 hinnangut Goodreads-ist)
(Director of Sciences at Yahoo Labs, Sunnyvale, CA, USA), (Staff applied researcher at LinkedIn Inc.)
  • Formaat: Paperback / softback, 264 pages, kõrgus x laius: 235x191 mm, kaal: 480 g
  • Ilmumisaeg: 26-Mar-2014
  • Kirjastus: Morgan Kaufmann Publishers In
  • ISBN-10: 0124071716
  • ISBN-13: 9780124071711
Teised raamatud teemal:
Relevance Ranking for Vertical Search EnginesSuurem pilt
  • Formaat: Paperback / softback, 264 pages, kõrgus x laius: 235x191 mm, kaal: 480 g
  • Ilmumisaeg: 26-Mar-2014
  • Kirjastus: Morgan Kaufmann Publishers In
  • ISBN-10: 0124071716
  • ISBN-13: 9780124071711
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9780124071711.html
Märksõnad:
Editors Long and Chang offer this guide to effective ranking of results in vertical searching, intended to be useful to researchers, graduate students, and systems engineers in the fields of information retrieval and data mining. The book starts off with learning and clustering algorithms for news and medical domain searching. Several categories of visual search are discussed including text-based, query example-based, and concept-based, as well as re-ranking paradigms. Mobile search ranking and entity ranking are also covered. The last several chapters address more complex aspects and applications of the ranking process, including striking the correct balance of multi-aspect relevance, aggregated vertical searching, and cross-vertical searching. Background in mathematics up to calculus and comfort with dense notation is helpful. Annotation ©2014 Ringgold, Inc., Portland, OR (protoview.com)

In plain, uncomplicated language, and using detailed examples to explain the key concepts, models, and algorithms in vertical search ranking,Relevance Ranking for Vertical Search Engines teaches readers how to manipulate ranking algorithms to achieve better results in real-world applications.

This reference book for professionals covers concepts and theories from the fundamental to the advanced, such as relevance, query intention, location-based relevance ranking, and cross-property ranking. It covers the most recent developments in vertical search ranking applications, such as freshness-based relevance theory for new search applications, location-based relevance theory for local search applications, and cross-property ranking theory for applications involving multiple verticals.

  • Foreword by Ron Brachman, Chief Scientist and Head, Yahoo! Labs
  • Introduces ranking algorithms and teaches readers how to manipulate ranking algorithms for the best results
  • Covers concepts and theories from the fundamental to the advanced
  • Discusses the state of the art: development of theories and practices in vertical search ranking applications
  • Includes detailed examples, case studies and real-world situations

Arvustused

"The authors of this book are active researchers in vertical search technology. This book provides researchers and application developers a comprehensive overview of the general concepts, techniques, and applications in vertical search." --Prabhakar Raghavan, Vice President of Engineering at Google

"This is an excellent book that gives the first and comprehensive introduction and overview on vertical search, an emerging and important field! Researchers and practitioners will find this book provides a comprehensive overview and systematic treatment on theories, methodologies and practices for vertical search ranking, covering several very promising topics, such as search on news and medical information, entity search, mobile search, as well as multi-aspect ranking, aggregating vertical search ranking and cross vertical ranking. I found it a great pleasure to read!" --Jiawei Han, Abel Bliss Professor, Department of Computer Science, Univ. of Illinois at Urbana-Champaign

Muu info

Learn in-depth and systematic practices including state-of-the-art algorithms to lay a solid foundation for future advance in the field of vertical search ranking.
List of Tables ix
List of Figures xiii
About the Editors xvii
Contributors xix
Foreword xxi
Chapter 1 Introduction 1(6)
1.1 Defining the Area
1(1)
1.2 The Content and Organization of This Book
1(4)
1.3 The Audience for This Book
5(1)
1.4 Further Reading
5(2)
Chapter 2 News Search Ranking 7(36)
2.1 The Learning-to-Rank Approach
7(3)
2.1.1 Related Works
8(1)
2.1.2 Combine Relevance and Freshness
8(2)
2.2 Joint Learning Approach from Clickthroughs
10(17)
2.2.1 Joint Relevance and Freshness Learning
12(2)
2.2.2 Temporal Features
14(3)
2.2.3 Experiment Results
17(2)
2.2.4 Analysis of JRFL
19(5)
2.2.5 Ranking Performance
24(3)
2.3 News Clustering
27(15)
2.3.1 Architecture of the System
29(1)
2.3.2 Offline Clustering
30(3)
2.3.3 Incremental Clustering
33(1)
2.3.4 Real-Time Clustering
34(3)
2.3.5 Experiments
37(5)
Summary
42(1)
Chapter 3 Medical Domain Search Ranking 43(16)
Introduction
43(1)
3.1 Search Engines for Electronic Health Records
44(3)
3.2 Search Behavior Analysis
47(2)
3.3 Relevance Ranking
49(5)
3.3.1 Insights from the TREC Medical Record Track
50(2)
3.3.2 Implementing and Evaluating Relevance Ranking in EHR Search Engines
52(2)
3.4 Collaborative Search
54(3)
3.5 Conclusion
57(2)
Chapter 4 Visual Search Ranking 59(22)
Introduction
59(1)
4.1 Generic Visual Search System
60(1)
4.2 Text-Based Search Ranking
61(3)
4.2.1 Text Search Models
61(1)
4.2.2 Textual Query Preprocessing
62(1)
4.2.3 Text Sources
63(1)
4.3 Query Example-Based Search Ranking
64(4)
4.3.1 Low-Level Visual Features
64(1)
4.3.2 Distance Metrics
65(3)
4.4 Concept-Based Search Ranking
68(3)
4.4.1 Query-Concept Mapping
68(2)
4.4.2 Search with Related Concepts
70(1)
4.5 Visual Search Reranking
71(5)
4.5.1 First Paradigm: Self-Reranking
71(2)
4.5.2 Second Paradigm: Example-Based Reranking
73(1)
4.5.3 Third Paradigm: Crowd Reranking
74(1)
4.5.4 Fourth Paradigm: Interactive Reranking
75(1)
4.6 Learning and Search Ranking
76(4)
4.6.1 Ranking by Classification
76(1)
4.6.2 Classification vs. Ranking
77(1)
4.6.3 Learning to Rank
78(2)
4.7 Conclusions and Future Challenges
80(1)
Chapter 5 Mobile Search Ranking 81(26)
Introduction
81(2)
5.1 Ranking Signals
83(4)
5.1.1 Distance
84(1)
5.1.2 Customer Reviews and Ratings
84(1)
5.1.3 Personal Preference
85(1)
5.1.4 Search Context: Location, Time, and Social Factors
85(2)
5.2 Ranking Heuristics
87(17)
5.2.1 Dataset and Experimental Setting
88(2)
5.2.2 Customer Rating
90(5)
5.2.3 Number of Reviews
95(1)
5.2.4 Distance
96(3)
5.2.5 Personal Preference
99(3)
5.2.6 Sensitivity Analysis
102(2)
5.3 Summary and Future Directions
104(3)
5.3.1 Evaluation of Mobile Local Search
104(1)
5.3.2 User Modeling and Personalized Search
105(2)
Chapter 6 Entity Ranking 107(20)
6.1 An Overview of Entity Ranking
107(2)
6.2 Background Knowledge
109(4)
6.2.1 Terminology
109(2)
6.2.2 Knowledge Base
111(1)
6.2.3 Web Search Experience
112(1)
6.3 Feature Space Analysis
113(3)
6.3.1 Probabilistic Feature Framework
113(2)
6.3.2 Graph-Based Entity Popularity Feature
115(1)
6.4 Machine-Learned Ranking for Entities
116(4)
6.4.1 Problem Definition
117(1)
6.4.2 Pairwise Comparison Model
117(2)
6.4.3 Training Ranking Function
119(1)
6.5 Experiments
120(5)
6.5.1 Experimental Setup
120(1)
6.5.2 User Data-Based Evaluation
121(3)
6.5.3 Editorial Evaluation
124(1)
6.6 Conclusions
125(2)
Chapter 7 Multi-Aspect Relevance Ranking 127(20)
Introduction
127(2)
7.1 Related Work
129(2)
7.2 Problem Formulation
131(4)
7.2.1 Learning to Rank for Vertical Searches
131(2)
7.2.2 Multi-Aspect Relevance Formulation
133(1)
7.2.3 Label Aggregation
133(1)
7.2.4 Model Aggregation
134(1)
7.3 Learning Aggregation Functions
135(3)
7.3.1 Learning Label Aggregation
135(2)
7.3.2 Learning Model Aggregation
137(1)
7.4 Experiments
138(7)
7.4.1 Datasets
138(2)
7.4.2 Ranking Algorithms
140(1)
7.4.3 Offline Experimental Results
141(2)
7.4.4 Online Experimental Results
143(2)
7.5 Conclusions and Future Work
145(2)
Chapter 8 Aggregated Vertical Search 147(34)
Introduction
147(2)
8.1 Sources of Evidence
149(9)
8.1.1 Types of Features
149(3)
8.1.2 Query Features
152(1)
8.1.3 Vertical Features
153(1)
8.1.4 Vertical-Query Features
154(4)
8.1.5 Implementation Details
158(1)
8.2 Combination of Evidence
158(8)
8.2.1 Vertical Selection
158(4)
8.2.2 Vertical Presentation
162(4)
8.3 Evaluation
166(10)
8.3.1 Vertical Selection Evaluation
167(1)
8.3.2 End-to-End Evaluation
168(8)
8.4 Special Topics
176(3)
8.4.1 Dealing with New Verticals
176(3)
8.4.2 Explore/Exploit
179(1)
8.5 Conclusion
179(2)
Chapter 9 Cross-Vertical Search Ranking 181(20)
Introduction
181(1)
9.1 The PCDF Model
182(4)
9.1.1 Problem Formulation
182(1)
9.1.2 Model Formulation
183(3)
9.2 Algorithm Derivation
186(5)
9.2.1 Objective Specification
187(2)
9.2.2 Optimization and Implementation
189(2)
9.3 Experimental Evaluation
191(7)
9.3.1 Data
192(1)
9.3.2 Experimental Setting
193(1)
9.3.3 Results and Discussions
193(5)
9.4 Related Work
198(2)
9.5 Conclusions
200(1)
References 201(22)
Author Index 223(10)
Subject Index 233
Bo Long is currently a staff applied researcher at LinkedIn Inc., and was formerly a senior research scientist at Yahoo! Labs. His research interests lie in data mining and machine learning with applications to web search, recommendation, and social network analysis. He holds eight innovations and has published peer-reviewed papers in top conferences and journals including ICML, KDD, ICDM, AAAI, SDM, CIKM, and KAIS. He has served as reviewer, workshop co-organizer, conference organizer, committee member, and area chair for multiple conferences, including KDD, NIPS, SIGIR, ICML, SDM, CIKM, JSM etc. Dr. Yi Chang is director of sciences in Yahoo Labs, where he leads the search and anti-abuse science group. His research interests include web search, applied machine learning, and social media mining. Yi has published more than 70 conference/journal papers, and he is a co-author of the book, Relevance Ranking for Vertical Search Engines. Yi is an associate editor for Neurocomputing, Pattern Recognition Letters, and he has served as workshops co-organizers, conference organizer committee members, and area chairs for multiple conferences, including WWW, SIGIR, ICML, KDD, CIKM, etc.