Proceedings of the 39th International ACM SIGIR Conference on Research and Development in Information Retrieval 2016
DOI: 10.1145/2911451.2911531
|View full text |Cite
|
Sign up to set email alerts
|

Learning Query and Document Relevance from a Web-scale Click Graph

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
29
0

Year Published

2017
2017
2023
2023

Publication Types

Select...
3
3
3

Relationship

0
9

Authors

Journals

citations
Cited by 46 publications
(29 citation statements)
references
References 25 publications
0
29
0
Order By: Relevance
“…We choose 256 as the dimensions of both word vectors and product vectors. • VPCG (Vector Propagation on Click Graphs) [18] is a stateof-the-art non-neural graph-based retrieval model that utilizes vector propagation on click graphs to overcome the above-mentioned long-tail problem and the lexical gap between queries and documents. • ARC-II [15], MatchPyramid [32] are two state-of-the-art neural retrieval models .…”
Section: Baselinesmentioning
confidence: 99%
See 1 more Smart Citation
“…We choose 256 as the dimensions of both word vectors and product vectors. • VPCG (Vector Propagation on Click Graphs) [18] is a stateof-the-art non-neural graph-based retrieval model that utilizes vector propagation on click graphs to overcome the above-mentioned long-tail problem and the lexical gap between queries and documents. • ARC-II [15], MatchPyramid [32] are two state-of-the-art neural retrieval models .…”
Section: Baselinesmentioning
confidence: 99%
“…Gao et al [11] propose a random walk-based smoothing approach to overcome the sparsity problem of clickthrough data in web search. Click graphs are also used to learn semantic representations of queries and documents (e.g., [18,46]) and to learn query intent (e.g., [23,25]). Ren et al [36] propose to use a heterogeneous graph that consists of queries, web pages, and Wikipedia concepts to improve search performance.…”
Section: Graph-based Information Retrievalmentioning
confidence: 99%
“…Another relevant approach that can be used to estimate termweights is to first construct a click graph between the queries and documents and estimate a vector representation for each entity (queries and documents) using a vector propagation model (VPCG & VG) [9]. The queries or documents can then be broken down into individual units (e.g., n-grams), and we can learn a vector representation for each n-gram based on the vectors already estimated from the click graph.…”
Section: Term-weighting Methodsmentioning
confidence: 99%
“…[43] explores various embedding estimations of the queries for a speci c topic extraction. [14] utilizes Web-click graphs to rank documents for a given query. Unseen Topic Detection Clustering, TFIDF, LDA [23] leverages TFIDF for document similarity to further lter and enhance event detection.…”
Section: Short Descriptionmentioning
confidence: 99%