Semi-subsumed Events: A Probabilistic Semantics of the BM25 Term Frequency Quantification

Wu, Hengzhi; Roelleke, Thomas

doi:10.1007/978-3-642-04417-5_43

Cited by 3 publications

(4 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, this led to the analytical description of what TF quantifications capture, namely the dependence between multiple event occurrences expressed via the generalized harmonic sum. This semantics is more general than the semantics presented in [25], where it was reported that the BM25-TF is an assumption between independence and subsumption. This argument is based on the fact that for independence the sequence probability is p n and for subsumption it is p 1 .…”

Section: Ir Modelsmentioning

confidence: 96%

“…The research regarding 'event spaces' [22] and dualities between IR models [23,24] produced a range of insights. This pathway led to the notion 'semi-subsumption' [25], an assumption lying between independence and subsumption.…”

Section: Harmony Assumptions In Information Retrieval and Social Networkmentioning

confidence: 99%

“…Gaussian harmony is the assumption that corresponds to the notion of semi-subsumed events [25], which was the first attempt to explain the dependence underlying the BM25-TF.…”

Section: Gaussian Harmonymentioning

confidence: 99%

See 2 more Smart Citations

Harmony Assumptions in Information Retrieval and Social Networks

Roelleke

Kaltenbrunner

Baeza-Yates

2015

The Computer Journal

Self Cite

View full text Add to dashboard Cite

In many applications, independence of event occurrences is assumed, even if there is evidence for dependence. Capturing dependence leads to complex models, and even if the complex models were superior, they fail to beat the simplicity and scalability of the independence assumption. Therefore, many models assume independence and apply heuristics to improve results. Theoretical explanations of the heuristics are seldom given or generalizable. This paper reports that some of these heuristics can be explained as encoding dependence in an exponent based on the generalized harmonic sum. Unlike independence, where the probability of subsequent occurrences of an event is the product of the single event probability, harmony is based on a product with decaying exponent. For independence, the sequence probability is p 1+1+•••+1 = p n , whereas for harmony, it is p 1+1/2+•••+1/n . The generalized harmonic sum leads to a spectrum of harmony assumptions. This paper shows that harmony assumptions naturally extend probability theory. An experimental evaluation for information retrieval (IR; term occurrences) and social networks (SN's; user interactions) shows that assuming harmony is more suitable than assuming independence. The potential impact of harmony assumptions lies beyond IR and SN's, since many applications rely on probability theory and apply heuristics to compensate the independence assumption. Given the concept of harmony assumptions, the dependence between multiple occurrences of an event can be reflected in an intuitive and effective way.

show abstract

Section: Ir Modelsmentioning

confidence: 96%

Section: Harmony Assumptions In Information Retrieval and Social Networkmentioning

confidence: 99%

See 1 more Smart Citation

Harmony Assumptions in Information Retrieval and Social Networks

Roelleke

Kaltenbrunner

Baeza-Yates

2015

The Computer Journal

Self Cite

View full text Add to dashboard Cite

show abstract

“…Moreover, the tutorial will review that LM is based on P (q|d)/P (q), and what makes the relationship between LM and the probability of relevance P (r|d, q) [9]. Finally, the tutorial will include conducive interpretations of the renown BM25 TF quantification tf/(tf+K) [3,26,19]. After the tutorial, the participants will have their view on statements such as "we know that LM works, but we do not know why"; "TF-IDF is intuitive, LM is not"; "TF-IDF is heuristic, whereas LM has a probabilistic semantics".…”

Section: Ir Models: Relationships (90 Mins)mentioning

confidence: 99%

IR models

Roelleke

2012

Proceedings of the 35th International ACM SIGIR Conference on Research and Development in Information Retrieval

Self Cite

View full text Add to dashboard Cite

In IR research it is essential to know IR models. Research over the past years has consolidated the foundations of IR models. Moreover, relationships have been reported that help to use and position IR models. Knowing about the foundations and relationships of IR models can significantly improve building information management systems. The first part of this tutorial presents an in-depth consolidation of the foundations of the main IR models (TF-IDF, BM25, LM). Particular attention will be given to notation and probabilistic roots. The second part crystallises the relationships between models. Does LM embody IDF? How "heuristic" is TF-IDF? What are the probabilistic roots? How are LM and the probability of relevance related? What are the components shared by the main IR models? After the tutorial, attendees will be familiar with a consolidated view on IR models. The tutorial will be illustrative and interactive, providing opportunities to exchange controversial issues and research challenges.

show abstract

Information Retrieval Models: Foundations and Relationships

Roelleke

2013

Synthesis Lectures on Information Concepts, Retrieval, and Serv

View full text Add to dashboard Cite

Semi-subsumed Events: A Probabilistic Semantics of the BM25 Term Frequency Quantification

Cited by 3 publications

References 3 publications

Harmony Assumptions in Information Retrieval and Social Networks

Harmony Assumptions in Information Retrieval and Social Networks

IR models

Information Retrieval Models: Foundations and Relationships

Contact Info

Product

Resources

About