Learning relevance from click data via neural network based similarity models

Ye, Xugang; Qi, Zijie; Massey, Dan

doi:10.1109/bigdata.2015.7363825

Cited by 5 publications

(6 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DSSM was one of the pioneering models incorporating click-through data in deep NNs. It has been built on by a variety of others (Mitra 2015;Mitra and Craswell 2015;Shen et al 2014a, b;Ye et al 2015). Other work in this category is either an architectural variant of DSSM with different SCNs or they propose novel ways of using distributed representations by DSSM variants in order to improve retrieval effectiveness.…”

Section: Learnmentioning

confidence: 99%

“…Li et al (2014) utilize distributed representations produced by DSSM and CLSM in order to re-rank documents based on in-session contextual information. Ye et al (2015) question the assumptions about clicked query-document pairs in order to derive triplets for training variants of the DSSM models.…”

Section: Learnmentioning

confidence: 99%

“…In cases where it is crucial to learn a semantic compositionality model with a CNN or RNN architecture, the large vocabulary brings a challenge. Word hashing, first introduced in Huang et al (2013) together with the DSSM model and adopted by Mitra (2015), Mitra and Craswell (2015), Shen et al (2014a, b), Ye et al (2015), reduces large vocabularies to a moderate sized vocabulary of 30K of character trigrams. Words are broken down into letter n-grams and then represented as a vector of letter n-grams.…”

Section: How To Cope With Large Vocabularies?mentioning

confidence: 99%

“…Strong empirical evidence is still needed to demonstrate that adding a KB does in indeed benefit the deep neural architecture for capturing semantic similarity. Ye et al (2015) question assumptions about clicked query-document pairs in order to derive triplets for training DSSM variant models. According to Ye et al (2015) these three assumptions are: (1) each clicked query-document pair is equally weighted; (2) each clicked query-document pair is a 100% positive example; and (3) each click is solely due to semantic similarity.…”

Section: Learnmentioning

confidence: 99%

“…Ye et al (2015) question assumptions about clicked query-document pairs in order to derive triplets for training DSSM variant models. According to Ye et al (2015) these three assumptions are: (1) each clicked query-document pair is equally weighted; (2) each clicked query-document pair is a 100% positive example; and (3) each click is solely due to semantic similarity. The authors relax these assumption and propose two generalized extensions to DSSM: GDSSM1 and GDSSM2.…”

Section: Learnmentioning

confidence: 99%

See 4 more Smart Citations

Neural information retrieval: at the end of the early years

et al. 2017

View full text Add to dashboard Cite

A recent ''third wave'' of neural network (NN) approaches now delivers state-ofthe-art performance in many machine learning tasks, spanning speech recognition, computer vision, and natural language processing. Because these modern NNs often comprise multiple interconnected layers, work in this area is often referred to as deep learning. Recent years have witnessed an explosive growth of research into NN-based approaches to information retrieval (IR). A significant body of work has now been created. In this paper,

show abstract

Section: Learnmentioning

confidence: 99%

Section: Learnmentioning

confidence: 99%

Section: How To Cope With Large Vocabularies?mentioning

confidence: 99%

Section: Learnmentioning

confidence: 99%

Section: Learnmentioning

confidence: 99%

See 3 more Smart Citations

Neural information retrieval: at the end of the early years

et al. 2017

View full text Add to dashboard Cite

show abstract

Information retrieval versus deep learning approaches for generating traceability links in bilingual projects

2021

View full text Add to dashboard Cite

Traceability Support for Multi-Lingual Software Projects

Liu

Lin

Cleland‐Huang

2020

Proceedings of the 17th International Conference on Mining Software Repositories

View full text Add to dashboard Cite

Software traceability establishes associations between diverse software artifacts such as requirements, design, code, and test cases. Due to the non-trivial costs of manually creating and maintaining links, many researchers have proposed automated approaches based on information retrieval techniques. However, many globally distributed software projects produce software artifacts written in two or more languages. The use of intermingled languages reduces the efficacy of automated tracing solutions. In this paper, we first analyze and discuss patterns of intermingled language use across multiple projects, and then evaluate several different tracing algorithms including the Vector Space Model (VSM), Latent Semantic Indexing (LSI), Latent Dirichlet Allocation (LDA), and various models that combine mono-and cross-lingual word embeddings with the Generative Vector Space Model (GVSM). Based on an analysis of 14 Chinese-English projects, our results show that best performance is achieved using mono-lingual word embeddings integrated into GVSM with machine translation as a preprocessing step.

show abstract

Learning relevance from click data via neural network based similarity models

Cited by 5 publications

References 12 publications

Neural information retrieval: at the end of the early years

Neural information retrieval: at the end of the early years

Information retrieval versus deep learning approaches for generating traceability links in bilingual projects

Traceability Support for Multi-Lingual Software Projects

Contact Info

Product

Resources

About