Information retrieval as statistical translation

By Adam L. Berger, John D. Lafferty. 1999

Subfields: Information Retrieval & Search

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Retrieval & Search

Citations in the corpus

End-to-End Neural Ad-hoc Ranking with Kernel Pooling (2017, methodology): “ey model query-document relevance using a pre-computed translation matrix that describes the similarities between word pairs [1].”
Pretrained Transformers for Text Ranking: BERT and Beyond (2020, methodology): “A notable attempt is the statistical translation approach of Berger and Lafferty [1999], who modeled retrieval as the translation of a document into a query in a noisy channel model.”
Modeling Interestingness with Deep Neural Networks (2014, positive): “WTM (Row 3) is our implementation of the word translation model for IR (Berger and Lafferty 1999; Gao et al. 2010).”
Document Expansion by Query Prediction (2019, background): “Approaches to tackling the vocabulary mismatch problem within these constraints include relevance feedback (Rocchio, 1971), query expansion (Voorhees, 1994; Xu and Croft, 2000), and modeling term relationships using statistical translation (Berger and Lafferty, 1999).”
Headline Generation Based on Statistical Translation (2000, background): “In a similar vein, a summarizer can be considered to be ‘translat-ing’ between two languages: one verbose and the other succinct (Berger and Lafferty, 1999; Wit-brock and Mittal, 1999).”
Context-Aware Document Term Weighting for Ad-Hoc Search (2020, methodology): “IR research mainly focuses on two types of weak relevance signals: content-based signals [1–3, 19, 25] and pseudo-relevance feedback based signals [13, 41, 42].”
Semantic term matching in axiomatic approaches to information retrieval (2006, methodology): “For example, co-occurrences of terms obtained from the analysis of a document collection usually reflect underlying semantic relationships that exist between terms [23, 3, 4, 2], and we may use measures such as Dice similarity [1] and mutual information [25, 15, 9, 8, 13, 7] to compute term similari”
Convolutional Neural Networks for Soft-Matching N-Grams in Ad-hoc Search (2018, unknown): “ey are rooted in statistical translation models, which construct a translation matrix of word pairs between query and document, and summarize it to a ranking score [4].”
A Noisy-Channel Approach to Question Answering (2003, background): “, 1993), information retrieval (Berger and Lafferty, 1999), and text summarization (Knight and Marcu, 2002), we develop a noisy channel model for QA.”
Semantic Matching in Search (2014, background): “Berger and Lafferty [22] propose formulating search as SMT problem, in which query q is translated into document d with the largest conditional”
Selective Weak Supervision for Neural Information Retrieval (2020, methodology): “Interaction based methods model the fine-grained interactions between query and documents, often by the translation matrices between all query and document term pairs [2], which in Neu-IR are calculated using term embeddings [18].”
Similarity Measures for Short Segments of Text (2007, unknown): “Many techniques have been proposed to overcome the vocabulary mismatch problem, including stemming [5] [9], LSI [3], translation models [1], and query expansion [6] [14].”
Two decades of statistical language modeling: where do we go from here? (2000, methodology): “In information retrieval, a language modeling approach was recently proposed by [3], and a statistical/information theoretical approach was developed by [4].”
Information Retrieval: A Biomedical and Health Perspective (2020, background): “One researcher found that the best reviews came from faculty of junior academic status [176], while another showed the best reviews came from younger faculty working at top academic institutions or who were known to the editors [177].”
Model-based feedback in the language modeling approach to information retrieval (2001, background): “In general, any approach to the retrieval problem is decomposed into three basic components: (1) query representation; (2) document representation; and (3) matching of query representation and document representation.”
The Turn - Integration of Information Seeking and Retrieval in Context (2005, methodology): “• Language models represent queries and documents directly as probabilistic models and integrate document indexing and retrieval into a single model – see Hiemstra (1998a-b), Ponte and Croft (1998), and Berger and Lafferty (1999). • Logical models: Van Rijsbergen (1986) proposed the Logical Uncertai”