Relevance-Ranked Domain-Specific Synonym Discovery

Yates, Andrew; Goharian, Nazli

doi:10.1007/978-3-319-06028-6_11

Cited by 7 publications

(6 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work we replicated the synonym discovery method proposed by Yates et al [25] in a new language (i.e., Swedish rather than English) and in a new domain (i.e., building construction rather than medical side effects). We found that in the new domain, the proposed LogReg method outperformed the PMI baseline as before.…”

Section: Discussionmentioning

confidence: 87%

“…Yates et al [25] proposed a method that produces a ranked list of domainspecific synonyms using a domain-specific corpus as input. Their learning to rank approach uses a set of features that outperformed previous synonym discovery methods that relied on single statistical measures: pointwise mutual information over term co-occurrences [23] and pointwise total correlation between two terms and the syntactic context (dependency relations) in which the terms appear [10].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Replicating Relevance-Ranked Synonym Discovery in a New Language and Domain

Yates

Unterkalmsteiner

2019

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Domain-specific synonyms occur in many specialized search tasks, such as when searching medical documents, legal documents, and software engineering artifacts. We replicate prior work on ranking domainspecific synonyms in the consumer health domain by applying the approach to a new language and domain: identifying Swedish language synonyms in the building construction domain. We chose this setting because identifying synonyms in this domain is helpful for downstream systems, where different users may query for documents (e.g., engineering requirements) using different terminology. We consider two new features inspired by the change in language and methodological advances since the prior work's publication. An evaluation using data from the building construction domain supports the finding from the prior work that synonym discovery is best approached as a learning to rank task in which a human editor views ranked synonym candidates in order to construct a domain-specific thesaurus. We additionally find that FastText embeddings alone provide a strong baseline, though they do not perform as well as the strongest learning to rank method. Finally, we analyze the performance of individual features and the differences in the domains.

show abstract

Section: Discussionmentioning

confidence: 87%

Section: Introductionmentioning

confidence: 99%

Replicating Relevance-Ranked Synonym Discovery in a New Language and Domain

Yates

Unterkalmsteiner

2019

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Document ranking in the broad medical domain have received extensive interest of researchers [87,165,174,178,189,225]. However, these efforts focus on conventional query-document retrieval.…”

Section: Significant Discrepancymentioning

confidence: 99%

Effective and practical neural ranking

MacAvaney

2021

SIGIR Forum

View full text Add to dashboard Cite

Supervised machine learning methods that use neural networks ("deep learning") have yielded substantial improvements to a multitude of Natural Language Processing (NLP) tasks in the past decade. Improvements to Information Retrieval (IR) tasks, such as ad-hoc search, lagged behind those in similar NLP tasks, despite considerable community efforts. Although there are several contributing factors, I argue in this dissertation that early attempts were not more successful because they did not properly consider the unique characteristics of IR tasks when designing and training ranking models. I first demonstrate this by showing how large-scale datasets containing weak relevance labels can successfully replace training on in-domain collections. This technique improves the variety of queries encountered when training and helps mitigate concerns of over-fitting particular test collections. I then show that dataset statistics available in specific IR tasks can be easily incorporated into neural ranking models alongside the textual features, resulting in more effective ranking models. I also demonstrate that contextualized representations, particularly those from transformer-based language models, considerably improve neural ad-hoc ranking performance. I find that this approach is neither limited to the task of ad-hoc ranking (as demonstrated by ranking clinical reports) nor English content (as shown by training effective cross-lingual neural rankers). These efforts demonstrate that neural approaches can be effective for ranking tasks. However, I observe that these techniques are impractical due to their high query-time computational costs. To overcome this, I study approaches for offloading computational cost to index-time, substantially reducing query-time latency. These techniques make neural methods practical for ranking tasks. Finally, I take a deep dive into better understanding the linguistic biases of the methods I propose compared to contemporary and traditional approaches. The findings from this analysis highlight potential pitfalls of recent methods and provide a way to measure progress in this area going forward.

show abstract

“…Viewing this problem as ranking is a more suitable and practical form of evaluation; given a doctor's limited time, it is important for them to be presented with the reports that have the most significant discrepancies. Document ranking in the broad medical domain have received extensive interest of researchers [21,13,17,16,8,15]. However, these efforts focus on conventional query-document retrieval.…”

Section: Significant Discrepancymentioning

confidence: 99%

Ranking Significant Discrepancies in Clinical Reports

MacAvaney

Cohan

Goharian

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Medical errors are a major public health concern and a leading cause of death worldwide. Many healthcare centers and hospitals use reporting systems where medical practitioners write a preliminary medical report and the report is later reviewed, revised, and finalized by a more experienced physician. The revisions range from stylistic to corrections of critical errors or misinterpretations of the case. Due to the large quantity of reports written daily, it is often difficult to manually and thoroughly review all the finalized reports to find such errors and learn from them. To address this challenge, we propose a novel ranking approach, consisting of textual and ontological overlaps between the preliminary and final versions of reports. The approach learns to rank the reports based on the degree of discrepancy between the versions. This allows medical practitioners to easily identify and learn from the reports in which their interpretation most substantially differed from that of the attending physician (who finalized the report). This is a crucial step towards uncovering potential errors and helping medical practitioners to learn from such errors, thus improving patient-care in the long run. We evaluate our model on a dataset of radiology reports and show that our approach outperforms both previously-proposed approaches and more recent language models by 4.5% to 15.4%.

show abstract

Relevance-Ranked Domain-Specific Synonym Discovery

Cited by 7 publications

References 30 publications

Replicating Relevance-Ranked Synonym Discovery in a New Language and Domain

Replicating Relevance-Ranked Synonym Discovery in a New Language and Domain

Effective and practical neural ranking

Ranking Significant Discrepancies in Clinical Reports

Contact Info

Product

Resources

About