Adaptive linear information retrieval models

Bollmann, P.; Wong, S. K. M.

doi:10.1145/42005.42023

Cited by 21 publications

(8 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Its popularity comes from the fact that queries and documents are represented both in the same vector notation, which suggests a geometrical interpretation of the retrieval process in an n-dimensional space. It suggests the use of many statistical techniques, such as association measures that were devel-oped in numerical taxonomy, and it supports various feedback techniques such as Rocchio's query feedback (Rocchio & Salton, 1965), or more perception-like feedback as in Bollmann and Wong (1987) and Wong, Yao, and Bollmann (1988). In the last decade, there have also been discussions about the underlying assumptions of this model.…”

Section: Introductionmentioning

confidence: 83%

On the necessity of term dependence in a query space for weighted retrieval

Bollmann-Sdorra

Raghavan

1998

J. Am. Soc. Inf. Sci.

View full text Add to dashboard Cite

In recent years, in the context of the vector space model, the view, held by many researchers, that documents, queries, terms, etc., are all elements of a common space has been challenged (Bollmann‐Sdorra & Raghavan, 1993). In particular, it was noted that term independence has to be investigated in the context of user preferences and it was shown, through counterexamples, that term independence can hold in the document space, but not in the query space and vice versa. In this article, we continue the investigation of query and document spaces with respect to the property of term independence. We prove, under realistic assumptions, that requiring term independence to hold in the query space is inconsistent with the goal of achieving better performance by means of weighted retrieval. The result that term independence in the query space is undesirable is obtained without making any assumption about whether or not the property of term independence holds in the document space. The results of this article reinforce our position that the properties of document and query spaces must be investigated separately, since the document and query spaces do not necessarily have the same properties.

show abstract

Section: Introductionmentioning

confidence: 83%

On the necessity of term dependence in a query space for weighted retrieval

Bollmann-Sdorra

Raghavan

1998

J. Am. Soc. Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…Decision theory [French 1986] has been applied to IR [Bollman and Wong 1987]. It was found to be the basis of the vector space model [Salton et al 1975;Wang et al 1992], its generalized version, Boolean model [Wong et al 1986], and a restricted form of its extended version [Wong et al 1989].…”

Section: Related Workmentioning

confidence: 99%

Building a framework for the probability ranking principle by a family of expected weighted rank

Dang

Luk

et al. 2009

ACM Trans. Inf. Syst.

View full text Add to dashboard Cite

A new principles framework is presented for retrieval evaluation of ranked outputs. It applies decision theory to model relevance decision preferences and shows that the Probability Ranking Principle (PRP) specifies optimal ranking. It has two new components, namely a probabilistic evaluation model and a general measure of retrieval effectiveness. Its probabilities may be interpreted as subjective or objective ones. Its performance measure is the expected weighted rank which is the weighted average rank of a retrieval list. Starting from this measure, the expected forward rank and some existing retrieval effectiveness measures (e.g., top n precision and discounted cumulative gain) are instantiated using suitable weighting schemes after making certain assumptions. The significance of these instantiations is that the ranking prescribed by PRP is shown to be optimal simultaneously for all these existing performance measures. In addition, the optimal expected weighted rank may be used to normalize the expected weighted rank of retrieval systems for (summary) performance comparison (across different topics) between systems. The framework also extends PRP and our evaluation model to handle graded relevance, thereby generalizing the discussed, existing measures (e.g., top n precision) and probabilistic retrieval models for graded relevance.

show abstract

“…Usefulness uses all preference relations [10], [11], [12], contained in each user's specification of the relative relevance of the items delivered. In the classical preference relation .…”

Section: Usefulness Measurementmentioning

confidence: 99%

Bilingual Search Engine Augmented with Query Expansion

Facundes¹,

Theva-Aksorn²

2007

Seventh International Conference on Intelligent Systems Design and Applications (ISDA 2007)

View full text Add to dashboard Cite

Nowadays, distributed information in the form of linguistic resources is accessible throughout the internet. Search engines have become crucial for information retrieval since users need to find and retrieve relevant information in various languages and forms. The objective of this paper is to investigate the application of query expansion technique to improve cross-language information retrieval in English and Thai. As the method of evaluation of query expansion, we investigate whether the expanded terms are useful for the search. Search CapabilitiesOperator Specification: Operator specification allows a user to logically relate multiple concepts together to define what information is needed. Our system supports queries with various combinations of Boolean operators AND, OR and NOT. Operators are processed from left to right.Query Expansion: The query expansion component employs the dictionary-based technique using part-of-speech information and semantic relations. We use LEXiTRON for this expansion process. When a user submits a query to the system, the query is expanded by adding the terms which are synonyms and related words. However, since a word can have many senses, the appropriate level of expansion must be determined. At the first level expanding, we retain every possible sense, i.e. every possible part-of-speech. In the second level, we perform a part-of-speech disambiguation. If the 1 stlevel expanded words include more than one part-ofspeech, only words that have the same part-of-speech will be selected to the 2 nd -level expanded word. Proper weight calculation is then performed on the resulting words from the 2 nd level. Query Translation:The queries are translated using the same dictionary as in the previous step.Similarity Measurement:The similarity measurement tells how alike two documents are, or how alike a document and a query are. To measure the

show abstract

Adaptive linear information retrieval models

Cited by 21 publications

References 2 publications

On the necessity of term dependence in a query space for weighted retrieval

On the necessity of term dependence in a query space for weighted retrieval

Building a framework for the probability ranking principle by a family of expected weighted rank

Bilingual Search Engine Augmented with Query Expansion

Contact Info

Product

Resources

About