Download Statistical language models for information retrieval by ChengXiang Zhai PDF

By ChengXiang Zhai

As on-line details grows dramatically, se's reminiscent of Google are enjoying a progressively more very important position in our lives. serious to all se's is the matter of designing an efficient retrieval version which can rank files effectively for a given question. This has been a important examine challenge in details retrieval for a number of a long time. some time past ten years, a brand new iteration of retrieval versions, sometimes called statistical language versions, has been effectively utilized to resolve many various info retrieval difficulties. in comparison with the normal types reminiscent of the vector area version, those new versions have a extra sound statistical origin and will leverage statistical estimation to optimize retrieval parameters. they could even be extra simply tailored to version non-traditional and intricate retrieval difficulties. Empirically, they generally tend to accomplish related or higher functionality than a conventional version with much less attempt on parameter tuning. This ebook systematically reports the massive physique of literature on utilising statistical language types to details retrieval with an emphasis at the underlying rules, empirically potent language types, and language types constructed for non-traditional retrieval projects. the entire correct literature has been synthesized to make it effortless for a reader to digest the study development completed to this point and notice the frontier of study during this region. The booklet additionally bargains practitioners an informative advent to a collection of essentially worthwhile language versions which could successfully remedy quite a few retrieval difficulties. No past wisdom approximately details retrieval is needed, yet a few uncomplicated wisdom approximately chance and data will be necessary for absolutely digesting all of the info. desk of Contents: advent / review of knowledge Retrieval versions / basic question chance Retrieval version / advanced question probability version / Probabilistic Distance Retrieval version / Language versions for precise Retrieval initiatives / Language versions for Latent subject research / Conclusions

Show description

Read Online or Download Statistical language models for information retrieval PDF

Best storage & retrieval books

Knowledge Representation and the Semantics of Natural Language

The ebook offers an interdisciplinary method of wisdom illustration and the remedy of semantic phenomena of common language, that is situated among man made intelligence, computational linguistics, and cognitive psychology. The proposed strategy relies on Multilayered prolonged Semantic Networks (MultiNets), which might be used for theoretical investigations into the semantics of average language, for cognitive modeling, for describing lexical entries in a computational lexicon, and for usual language processing (NLP).

Web data mining: Exploring hyperlinks, contents, and usage data

Internet mining goals to find helpful details and data from net links, web page contents, and utilization facts. even though net mining makes use of many traditional info mining thoughts, it's not only an program of conventional facts mining end result of the semi-structured and unstructured nature of the internet info.

Semantic Models for Multimedia Database Searching and Browsing

Semantic types for Multimedia Database looking and skimming starts off with the advent of multimedia details purposes, the necessity for the improvement of the multimedia database administration structures (MDBMSs), and the real concerns and demanding situations of multimedia structures. The temporal kinfolk, the spatial kin, the spatio-temporal family, and several other semantic versions for multimedia details structures also are brought.

Enterprise Content Management in Information Systems Research: Foundations, Methods and Cases

This ebook collects ECM study from the tutorial self-discipline of data structures and comparable fields to help teachers and practitioners who're attracted to figuring out the layout, use and impression of ECM platforms. It additionally offers a worthy source for college students and academics within the box. “Enterprise content material administration in info structures examine – Foundations, tools and situations” consolidates our present wisdom on how today’s agencies can deal with their electronic info resources.

Additional resources for Statistical language models for information retrieval

Sample text

Such a formulation is limited because we would not be able to model the redundancy among search results. A more general way of framing the retrieval problem is to take it as a decision problem in which a system would respond to a query by choosing a set of documents from a collection and presenting the documents in a certain way. Such a decision-theoretic view of retrieval has been formalized with Bayesian decision theory in [1, 89, 90], resulting in a general risk minimization framework for information retrieval.

Another effort to improve document representation involves introducing the term frequency directly into the model by using a multiple 2-Poisson mixture representation of documents [64]. The relationship between different event models of the document-generation model is discusses in [206]. In general, with examples of relevant and nonrelevant documents, we can easily estimate the parameters in a document-generation model. Specifically, in the document-generation model, we need to estimate two component models p(D|Q, r) and p(D|Q, r).

We naturally use language models to model documents and queries. Formally, let θQ denote the parameters of a query model, and let θD denote the parameters of a document model. A user U generates a query by first selecting θQ , according to a distribution p(θQ | U ). Using this model, a query Q is then generated with probability p(Q | θQ ). Similarly, the source selects a document model θD according to a distribution p(θD | S ), and then uses this model to generate a document D according to p(D | θD ).

Download PDF sample

Rated 4.91 of 5 – based on 47 votes