Entity Linking in Queries

Hasibi, Faegheh; Balog, Krisztian; Bratsberg, Svein Erik

doi:10.1145/2808194.2809473

Cited by 39 publications

(14 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another interesting work is that presented by Hasibi et al [25] (a follow-up of the NTNU-UiS entity-annotator presented at the ERD Challenge), where the authors describe a method to solve the query-level entity-linking problem. The method is based on three phases: (i) candidate annotations are generated aiming at maximum recall, by exploiting two sources: DBpedia and Google's Freebase Annotations of the ClueWeb Corpora (FACC); (ii) candidate annotations are assigned a score by combining, via a generative model, the Mixture of Language Models (MLM) with a commonness score; (iii) interpretations are iteratively generated by picking non-overlapping annotations, starting from the ones with higher score.…”

Section: Related Workmentioning

confidence: 99%

“…This entityannotator outperforms TagMe, which is used as a baseline. Hasibi et al [25] address the problem of semantic mapping, i.e., finding the ranked list of pertinent entities, possibly without explicit mentions in the query (for example, Ann_Dunham for query obama mother). In a follow up [26], they propose an annotator that employs supervised learning for the entity ranking step while tackling disambiguation with an unsupervised algorithm.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Smaph

Cornolti

Ferragina

Ciaramita

et al. 2018

ACM Trans. Inf. Syst.

View full text Add to dashboard Cite

We study the problem of linking the terms of a web-search query to a semantic representation given by the set of entities (a.k.a. concepts) mentioned in it. We introduce SMAPH, a system that performs this task using the information coming from a web search engine, an approach we call "piggybacking." We employ search engines to alleviate the noise and irregularities that characterize the language of queries. Snippets returned as search results also provide a context for the query that makes it easier to disambiguate the meaning of the query. From the search results, SMAPH builds a set of candidate entities with high coverage. This set is filtered by linking back the candidate entities to the terms occurring in the input query, ensuring high precision. A greedy disambiguation algorithm performs this filtering; it maximizes the coherence of the solution by iteratively discovering the pertinent entities mentioned in the query. We propose three versions of SMAPH that outperform state-of-the-art solutions on the known benchmarks and on the GERDAQ dataset, a novel dataset that we have built specifically for this problem via crowd-sourcing and that we make publicly available.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Smaph

Cornolti

Ferragina

Ciaramita

et al. 2018

ACM Trans. Inf. Syst.

View full text Add to dashboard Cite

show abstract

“…What is more, even methods that are designed in particular for short text perform significantly worse on queries than those that are specifically devised for queries [19]. • Another fundamental difference is that when documents are annotated with entities, "it is implicitly assumed that the text provides enough context for each entity mention to be resolved unambiguously" [32]. For queries, on the other hand, there is only limited context, or none at all.…”

Section: Entity Linking In Queriesmentioning

confidence: 99%

“…Semantic linking refers to the task of identifying entities "that are intended or implied by the user issuing the query" [53]. This problem was introduced as query mapping by Meij et al [53] and is also known as semantic mapping [32] and as (ranked) concepts to Wikipedia [18]. As the name we have adopted suggests, we seek to find entities that are semantically related to the query.…”

Section: Semantic Linkingmentioning

confidence: 99%

See 1 more Smart Citation

Understanding Information Needs

Balog

2018

Entity-Oriented Search

Self Cite

View full text Add to dashboard Cite

Understanding what the user is looking for is at the heart of delivering a quality search experience. After all, it is rather difficult to serve good results, unless we can comprehend the intent and meaning behind the user's query. Query understanding is the first step that takes place before the scoring of results. Its overall aim is to infer a semantically enriched representation of the information need. This involves, among others, classifying the query according to higher-level goals or intent, segmenting it into parts that belong together, interpreting the query structure, recognizing and disambiguating the mentioned entities, and determining if specific services or verticals 1 should be invoked. Such semantic analysis of queries has been a longstanding research area in information retrieval. In Sect. 7.1, we give a brief overview of IR approaches to query understanding. In the rest of the chapter, we direct our focus of attention to representing information needs with the help of structured knowledge repositories. The catchphrase "things, not strings" was coined by Google when introducing their Knowledge Graph. 2 It aptly describes the current chapter's focus: Capturing what the query is about by automatically annotating it with entries from a knowledge repository. These semantic annotations can then be utilized in downstream processing for result ranking (see Chap. 4) and/or result presentation. Specifically, in Sect. 7.2, we seek to identify the types or categories of entities that are targeted by the query. In Sect. 7.3, we perform entity linking in queries, which is about recognizing specific entity mentions and annotating them with unique identifiers from the underlying knowledge repository. Additionally, we consider the case of unresolvable ambiguity, when queries have multiple possible interpretations. 1 A vertical is a specific segment of online content. Some of the most common verticals include shopping, travel, job search, the automotive industry, medical information, and scholarly literature.

show abstract

Entity Linking in Enterprise Search: Combining Textual and Structural Information

Bhatia¹

2018

Unsupervised and Semi-Supervised Learning

View full text Add to dashboard Cite

Entity Linking in Queries

Cited by 39 publications

References 33 publications

Smaph

Smaph

Understanding Information Needs

Entity Linking in Enterprise Search: Combining Textual and Structural Information

Contact Info

Product

Resources

About