Abstract:The initial retrieval stage of information retrieval aims to generate as many relevant candidate documents as possible in a simple yet efficient way. Traditional term based retrieval methods like BM25 deal with the problem based on Bag-of-Words (BoW) representation, thus they only focus on exact matching (i.e., syntactic) and lack the consideration for semantically related words. That causes the typical vocabulary mismatch problem and the reduction of performance in terms of recall. The advance of distributed representation (i.e., embedding) of words and documents provides an efficient way to measure the semantic relevance between words. Since embedding can alleviate the vocabulary mismatch problem, it is suitable for the initial retrieval task. We conduct several experiments to compare term basedmodels with embedding basedmodels in terms of recall. We compare above two branches of the initial retrieval models on three representative retrieval tasks (Web-QA, Ad-hoc retrieval and CQA respectively). The results show that embedding based method and term based method are complementary for each other and higher recall can be achieved by combining the above two types of models based on scores or ranking position. We find that combination of the two types of the models based on ranking position usually perform better than combination based on score. Furthermore, since queries and documents are in different forms for diverse application scenarios, it can be observed that the relative performance of the two types are almost same but the absolute performance are significant different regarding to distinct scenarios.

Exploration of Term Dependence in Sentence Retrieval.

Exploration of Term Relationship for Bayesian Network Based Sentence Retrieval

MRF Based Approach for Sentence Retrieval.

Exploiting Semantic Knowledge Base for Patent Retrieval

A Semantic Query Expansion-Based Patent Retrieval Approach

Exploiting Syntactic and Semantic Relationships Between Terms for Opinion Retrieval

Type-Based Query Expansion for Sentence Retrieval

Bayesian Network Based Sentence Retrieval Model.

A Novel Dependency Language Model for Information Retrieval

ON SENTENCE-BASED INFORMATION RETRIEVAL MODEL

Term Dependence on the Semantic Web

Understanding and Improving Neural Ranking Models from a Term Dependence View.

A Comparison Between Term-Based and Embedding-Based Methods for Initial Retrieval

Selecting Expansion Terms As a Set Via Integer Linear Programming

Semantic Query Expansion and Context-Based Discriminative Term Modeling for Spoken Document Retrieval

Using Proximity in Query Focused Multi-document Extractive Summarization

Improved Semantic Retrieval of Spoken Content by Document/Query Expansion with Random Walk Over Acoustic Similarity Graphs

Semantic Aspect Retrieval for Encyclopedia.

Bagging to find better expansion words

Incorporating Semantic Knowledge with MRF Term Dependency Model in Medical Document Retrieval.

Exploration of Query Context for Information Retrieval