A Markov random field model for term dependencies

By Donald Metzler, W. Bruce Croft. 2005

1,003 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Natural Language Processing, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Artificial Intelligence

In Information Systems

In Natural Language Processing

In Probabilistic & Bayesian Methods

Citations in the corpus

Pretrained Transformers for Text Ranking: BERT and Beyond (2020, methodology): “Different aggregation techniques are compared against two baselines: SDM, row (2), refers to the sequential dependence model [Metzler and Croft, 2005].”
Context-Aware Sentence/Passage Term Importance Estimation For First Stage Retrieval (2019, background): “BOW and SDM stand for bag-of-word queries and sequential dependency queries [17]. tf is the classic query term frequency weights.”
Learning to Match using Local and Distributed Representations of Text for Web Search (2016, methodology): “Match proximity serves as a foundation for effective methods such as sequential dependence models [23].”
Learning to Reweight Terms with Distributed Representations (2015, methodology): “A regularized linear regression problem from the feature vectors onto term weights is formulated and the predicted term weights are used for both bag-of-words queries and term dependency queries.”
Large Language Models for Information Retrieval: A Survey (2023, neutral): “Notable examples of such methods include RM3 [83], LCE [84], KL expansion [85], and relevance modeling [86].”
Complementing Lexical Retrieval with Semantic Residual Embedding (2020, methodology): “Traditionally, first-stage retrieval has relied on term weighting models such as BM25 (Robertson and Walker, 1994), or query likelihood (Lafferty and Zhai, 2001), and have explored extensions such as n-grams (Metzler and Croft, 2005), controlled vocabularies (Rajashekar and Croft, 1995), and query e”
Discovering key concepts in verbose queries (2008, methodology): “In addition, we compare the effectiveness of this method to retrieval using sequential dependency model [22], denoted SeqDep<desc>, which integrates the base <desc> query with sequential bi-grams derived from query words, and uses ordered and un-ordered proximity operators.”
Semantic Matching by Non-Linear Word Transportation for Information Retrieval (2016, methodology): “Note that the parameters for each model are tuned with the 5-fold cross-validation method mentioned above using the typical ranges suggested by the original papers [32, 27, 1, 22, 20, 31, 30, 12].”
Training Curricula for Open Domain Answer Re-Ranking (2020, methodology): “To put performance of the neural rankers in context, we include the ranking effectiveness of Anserini’s [38] implementation of BM25 and SDM [25], both with default parameters, tuned on the validation set (‘Tuned’), and tuned on the test set (representing the optimal settings of parameters for this…”
Latent entity space: a novel retrieval approach for entity-bearing queries (2015, methodology): “We use all of their four reported runs as baselines: • SDM : Sequential dependence model Metzler and Croft (2005).”
Learning semantic representations using convolutional neural networks for web search (2014, background): “Modeling contextual information in search queries and documents is a long-standing research topic in information retrieval (IR) [2][4][8].”
From Neural Re-Ranking to Neural Ranking: Learning a Sparse Representation for Inverted Indexing (2018, neutral): “ACMReference Format: Hamed Zamani, Mostafa Dehghani, W. Bruce Croft, Erik Learned-Miller, and Jaap Kamps.”
CEDR: Contextualized Embeddings for Document Ranking (2019, methodology): “7 We do not report SDM performance on WebTrack due to its high cost of retrieval on the large ClueWeb collections.”
Learning to efficiently rank (2010, methodology): “The other is the less eﬃcient, but more eﬀective sequential dependence model [16] (SD).”
Yahoo! Learning to Rank Challenge Overview (2010, background): “Finally, there are also proximity features which try to quantify how far in the document are the query terms (the closer the better) (Metzler and Croft, 2005).”
Of Ivory and Smurfs: Loxodontan MapReduce Experiments for Web Search (2009, methodology): “The system we have developed is called Ivory, which integrates Metzler’s SMRF (Search using Markov Random Fields) retrieval engine [14, 13].”
Improvements that don't add up: ad-hoc retrieval results since 1998 (2009, background): “…windows, with a maximum size of four times the number of terms being scored, for every sequence of two or three terms in the original query [Metzler and Croft, 2005] (This diverges slightly from the original method. described in the paper, but the number of possible combinations grows…”