Learning Unsupervised Knowledge-Enhanced Representations to Reduce the Semantic Gap in Information Retrieval

Agosti, Maristella; Marchesin, Stefano; Silvello, Gianmaria

doi:10.1145/3417996

Cited by 18 publications

(16 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Scientific Literature document set for TREC PM 2017 (PM17) and 2018 (PM18) consists of a set of 26, 759, 399 MEDLINE abstracts, plus two additional sets: (i) 37, 007 abstracts from recent proceedings of the American Society of Clinical Oncology (ASCO), 8 and (ii) 33, 018 abstracts from recent proceedings of the American Association for Cancer Research (AACR). 9 These additional sets were added to increase the set of potentially relevant treatment information. Indeed, relevant literature articles can guide precision oncologists to the best-known treatment options for the patient's condition.…”

Section: Scientific Literaturementioning

confidence: 99%

“…Otherwise, to use pyndri, we should first have had indexed the collections using Indri [215] -which is built in C++. 9 As in the original paper, we remove stopwords using the Indri stoplist, 10 and we do not perform stemming.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…We adopt the models optimized for document retrieval and we keep Indri default values for RM3, that is R 1 = 10, m = 10, and α = 0.5. 9 We consider different categories of retrieval models in the first round of retrieval to evaluate their effectiveness in reducing the semantic gap. Precisely, we investigate whether models that are specifically designed to address the semantic gap retrieve relevant documents that lexical models fail to discover.…”

Section: Expansion Models Setupmentioning

confidence: 99%

“…To overcome the above limitations, we propose the Semantic-Aware Neural Framework for IR (SAFIR) [9], an unsupervised knowledge-enhanced neural framework for IR. SAFIR learns representations that are optimized for IR and encodes linguistic features relevant to address the semantic gap between queries and documents.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Developing unsupervised knowledge-enhanced models to reduce the semantic gap in information retrieval

Marchesin

2021

SIGIR Forum

Self Cite

View full text Add to dashboard Cite

In this thesis we tackle the semantic gap, a long-standing problem in Information Retrieval (IR). The semantic gap can be described as the mismatch between users' queries and the way retrieval models answer to such queries. Two main lines of work have emerged over the years to bridge the semantic gap: (i) the use of external knowledge resources to enhance the bag-of-words representations used by lexical models, and (ii) the use of semantic models to perform matching between the latent representations of queries and documents. To deal with this issue, we first perform an in-depth evaluation of lexical and semantic models through different analyses [Marchesin et al., 2019]. The objective of this evaluation is to understand what features lexical and semantic models share, if their signals are complementary, and how they can be combined to effectively address the semantic gap. In particular, the evaluation focuses on (semantic) neural models and their critical aspects. Each analysis brings a different perspective in the understanding of semantic models and their relation with lexical models. The outcomes of this evaluation highlight the differences between lexical and semantic signals, and the need to combine them at the early stages of the IR pipeline to effectively address the semantic gap. Then, we build on the insights of this evaluation to develop lexical and semantic models addressing the semantic gap. Specifically, we develop unsupervised models that integrate knowledge from external resources, and we evaluate them for the medical domain - a domain with a high social value, where the semantic gap is prominent, and the large presence of authoritative knowledge resources allows us to explore effective ways to address it. For lexical models, we investigate how - and to what extent - concepts and relations stored within knowledge resources can be integrated in query representations to improve the effectiveness of lexical models. Thus, we propose and evaluate several knowledge-based query expansion and reduction techniques [Agosti et al., 2018, 2019; Di Nunzio et al., 2019]. These query reformulations are used to increase the probability of retrieving relevant documents by adding to or removing from the original query highly specific terms. The experimental analyses on different test collections for Precision Medicine - a particular use case of Clinical Decision Support (CDS) - show the effectiveness of the proposed query reformulations. In particular, a specific subset of query reformulations allow lexical models to achieve top performing results in all the considered collections. Regarding semantic models, we first analyze the limitations of the knowledge-enhanced neural models presented in the literature. Then, to overcome these limitations, we propose SAFIR [Agosti et al., 2020], an unsupervised knowledge-enhanced neural framework for IR. SAFIR integrates external knowledge in the learning process of neural IR models and it does not require labeled data for training. Thus, the representations learned within this framework are optimized for IR and encode linguistic features that are relevant to address the semantic gap. The evaluation on different test collections for CDS demonstrate the effectiveness of SAFIR when used to perform retrieval over the entire document collection or to retrieve documents for Pseudo Relevance Feedback (PRF) methods - that is, when it is used at the early stages of the IR pipeline. In particular, the quantitative and qualitative analyses highlight the ability of SAFIR to retrieve relevant documents affected by the semantic gap, as well as the effectiveness of combining lexical and semantic models at the early stages of the IR pipeline - where the complementary signals they provide can be used to obtain better answers to semantically hard queries.

show abstract

Section: Scientific Literaturementioning

confidence: 99%

Section: Implementation Detailsmentioning

confidence: 99%

Section: Expansion Models Setupmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Developing unsupervised knowledge-enhanced models to reduce the semantic gap in information retrieval

Marchesin

2021

SIGIR Forum

Self Cite

View full text Add to dashboard Cite

show abstract

“…The AraPlagDet corpus was used to evaluate their work using precision, recall, and F-score metrics, and achieved 82.4 percent precision, 93.2 percent recall and 87.5 percent F-score. Agosti et al (2020) proposed an unsupervised neural framework for information retrieval. Semantic indexing, synonymy, and polysemy were used to eliminate semantic gaps, indicating a mismatch between document terms and queries.…”

Section: Related Workmentioning

confidence: 99%

A Discourse-Based Information Retrieval for Tamil Literary Texts

Ramalingam

Navaneethakrish

2021

Journal of Information and Communication Technology

View full text Add to dashboard Cite

Tamil literature has many valuable thoughts that can help the human community to lead a successful and a happy life. Tamil literary works are abundantly available and searched on the World Wide Web (WWW), but the existing search systems follow a keyword-based match strategy which fails to satisfy the user needs. This necessitates the demand for a focused Information Retrieval System that semantically analyses the Tamil literary text which will eventually improve the search system performance. This paper proposes a novel Information Retrieval framework that uses discourse processing techniques which aids in semantic analysis and representation of the Tamil Literary text. The proposed framework has been tested using two ancient literary works, the Thirukkural and Naladiyar, which were written during 300 BCE. The Thirukkural comprises 1330 couplets, each 7 words long, while the Naladiyar consists of 400 quatrains, each 15 words long. The proposed system, tested with all the 1330 Thirukkural couplets and 400 Naladiyar quatrains, achieved a mean average precision (MAP) score of 89%. The performance of the proposed framework has been compared with Google Tamil search and a keyword-based search which is a substandard version of the proposed framework. Google Tamil search achieved a MAP score of 56% and keyword-based method achieved a MAP score of 62% which shows that the discourse processing techniques improves the search performance of an Information Retrieval system.

show abstract