This paper presents our novel relevance feedback (RF) algorithm that uses the probabilistic document-context based retrieval model with limited relevance judgments for document re-ranking. Probabilities of the document-context based retrieval model are estimated from the top N (=20) documents in the initial retrieval. We use document-context based cosine similarity measure to find similar data for better probability estimation in order to reduce the data scarcity problem and the negative weighting problem. Our RF algorithm is promising because its mean average precision is statistically significantly better than the baseline using TREC-6 and TREC-7 data collections.

The vector space model (VSM) is one of the most widely used information retrieval (IR) models in both academia and industry. It was less effective at the Chinese ad hoc retrieval tasks than other retrieval models in the NTCIR-3 evaluation workshop, but comparable to those in the NTCIR-4 and NTCIR-5 workshops. We do not know whether the lower level performance was due to the VSM's inherent deficiencies or to a less effective normalization of document length. Hence we evaluated the VSM with various pivoted normalizations of document length using the NTCIR-3 collection for confirmation. We found that VSM's retrieval effectiveness with pivoted normalization was comparable to other competitive retrieval models (for example, 2-Poisson), and that VSM's retrieval speed with pivoted normalization was similar to competitive retrieval models (2-Poisson). We proposed a novel adaptive scheme that automatically estimates the (near) best parameters for pivoted document-length normalization based on query size; the new normalization is called adaptive pivoted document-length normalization . This scheme achieved good retrieval effectiveness, sometimes for short (title) queries and sometimes for long queries, without manually adjusting parameter values. We found that unique, adaptive pivoted normalization can enhance fixed pivoted normalizations for different test collections (TREC-5 and TREC-6). We also evaluated the VSM with the adaptive pivoted normalization using the pseudo-relevance feedback (PRF) and found that this type of VSM performs similarly to the competitive retrieval models (2-Poisson) with PRF. Hence, we conclude that the VSM with unique (adaptive) pivoted document-length normalization is effective for Chinese IR and that its retrieval effectiveness is comparable to that of other competitive retrieval models with or without PRF for the reference test collections used in this evaluation.

We propose a novel probabilistic retrieval model which weights terms according to their contexts in documents. The term weighting function of our model is similar to the language model and the binary independence model. The retrospective experiments (i.e., relevance information is present) illustrate the potential of our probabilistic context-based retrieval where the precision at the top 30 documents is about 43% for TREC-6 data and 52% for TREC-7 data.

We report on Korean monolingual, Chinese-Korean English-as-pivot bilingual, and Chinese-English bilingual CLIR experiments using MT software augmented with Web-based entity-oriented translation as resources in the NTCIR-4 environment. Simple stemming is helpful in improving bigram indexing for Korean retrieval. For word indexing, keeping nouns only is preferable. Web-based translation reduces untranslated terms left over after MT and substantially improves CLIR results. Translation concatenation is found to consistently improve CLIR effectiveness, while combining a retrieval list from bigram and word indexing is also helpful. A method to disambiguate multiple MT outputs using a log likelihood ratio threshold was tested. Depending on the nature of the title or description queries, bigram only or a retrieval combination, or relaxed or rigid evaluations, direct bilingual CLIR returned an average precision of 71--79% (English-Korean) and 76--84% (Chinese-English) of the corresponding Korean-Korean and English-English monolingual results. Using English as a pivot in Chinese-Korean CLIR provides about 55--65% the effectiveness that Korean alone does. Entity/terminology translation at the pivot language stage accounts for a large portion of this deficiency. A topic with comparatively worse Chinese-English bilingual result does not necessarily mean that it will continue to under-perform (after further transitive Korean translation) at the Korean retrieval level.

Queries have specific properties, and may need individualized methods and parameters to optimize retrieval. Length is one property. We look at how two-word queries may attain higher precision by re-ranking using word co-occurrence evidence in retrieved documents. Co-occurrence within document context is not sufficient, but window context including sentence context evidence can provide precision improvements at low recall region of 4 to 10% using initial retrieval results, and positively affects pseudo-relevance feedback.

With the advent of the Internet and intranets, substantial interest is being shown in Asian language information retrieval; especially in Chinese, which is a good example of an Asian ideographic language (other examples include Japanese and Korean). Since, in this type of language, spaces do not delimit words, an important issue is which index terms should be extracted from documents. This issue also has wider implications for indexing other languages such as agglutinating languages (e.g., Finnish and Turkish), archaic ideographic languages like Egyptian hieroglyphs, and other types of information such as data stored in genomic databases. Although comparisons of indexing strategies for Chinese documents have been made, almost all of them are based on a single retrieval model. This article compares the performance of various combinations of indexing strategies (i.e., character, word, short-word, bigram, and Pircs indexing) and retrieval models (i.e., vector space, 2-Poisson, logistic regression, and Pircs models). We determine which model (and its parameters) achieves the (near) best retrieval effectiveness without relevance feedback, and compare it with the open evaluations (i.e., TREC and NTCIR) for both long and title queries. In addition, we describe a more extensive investigation of retrieval efficiency. In particular, the storage cost of word indexing is only slightly more than character indexing, and bigram indexing is about double the storage cost of other indexing strategies. The retrieval time typically varies linearly with the number of unique terms in the query, which is supported by correlation values above 90%. The Pircs retrieval system achieves robust and good retrieval performance, but it appears to be the slowest method, whereas vector space models were not very effective in retrieval, but were able to respond quickly. For robust, near-best retrieval effectiveness, without considering storage overhead, the 2-Poisson model using bigram indexing appears to be a good compromise between retrieval effectiveness and efficiency for both long and title queries.

For information retrieval in the Chinese language, three representation methods for texts are popular, namely: 1-gram or character, bigram, and short-word. Each has its advantages as well as drawbacks. Employing more than one method may combine advantages from them and enhance retrieval effectiveness. We investigated two ways of using them simultaneously: mixing representations in documents and queries, and combining retrieval lists obtained via different representations. The experiments were done with the 170 MB evaluated Chinese corpora and 54 long and short queries available from the (TREC) program and using our Probabilistic Indexing and Retrieval Components System (PIRCS retrieval system). Experiments show that good retrieval need not depend on accurate word segmentation; approximate segmentation into short-words will do. Results also show and confirm that bigram representation alone works well; mixing characters with bigram representation boosts effectiveness further, but it is preferable to mix characters with short-word indexing which is more efficient, needs less resource, and gives better retrieval more often. Combining retrieval lists from short-word with character representation and from bigram indexing provides the best retrieval results but also at a substantial cost.

Article Improving two-stage ad-hoc retrieval for short queries Share on Authors: K. L. Kwok Computer Science Department, Queens College, CUNY, Flushing, NY Computer Science Department, Queens College, CUNY, Flushing, NYView Profile , M. Chan Computer Science Department, Queens College, CUNY, Flushing, NY Computer Science Department, Queens College, CUNY, Flushing, NYView Profile Authors Info & Claims SIGIR '98: Proceedings of the 21st annual international ACM SIGIR conference on Research and development in information retrievalAugust 1998 Pages 250–256https://doi.org/10.1145/290941.291003Online:01 August 1998Publication History 40citation553DownloadsMetricsTotal Citations40Total Downloads553Last 12 Months13Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

Three representation methods are empirically investigated for Chinese information retrieval: 1-gram (single character), bigram (two contiguous overlapping characters), and short-word indexing based on a simple segmentation of the text. The retrieval collection is the approximately 170 MB TREC-5 Chinese corpus of news articles, and 28 queries that are long and rich in wordings. Evaluation shows that 1-gram indexing is good but not sufficiently competitive, while bigram indexing works surprisingly well. Bigram indexing leads to a large index term space, three times that of short-word indexing, but is as good as short-word indexing in precision, and about 5% better in relevants retrieved. The best average non-interpolated precision is about 0.45, 17% better than 1-gram indexing and quite high for a mainly statistical approach. 1. Introduction While information retrieval (IR) in English has over thirty years of history, IR in Chinese is relatively recent. It is well-known that written Chi...

Article Free Access Share on A new method of weighting query terms for ad-hoc retrieval Author: K. L. Kwok Computer Science Dept., Queens College, City University of New York, Flushing, NY Computer Science Dept., Queens College, City University of New York, Flushing, NYView Profile Authors Info & Claims SIGIR '96: Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrievalAugust 1996 Pages 187–195https://doi.org/10.1145/243199.243266Published:18 August 1996Publication History 40citation891DownloadsMetricsTotal Citations40Total Downloads891Last 12 Months23Last 6 weeks3 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

Article Query modification and expansion in a network with adaptive architecture Share on Author: K. L. Kwok Computer Science Department, Queens College, CUNY, Flushing, NY Computer Science Department, Queens College, CUNY, Flushing, NYView Profile Authors Info & Claims SIGIR '91: Proceedings of the 14th annual international ACM SIGIR conference on Research and development in information retrievalSeptember 1991 Pages 192–201https://doi.org/10.1145/122860.122879Online:01 September 1991Publication History 6citation320DownloadsMetricsTotal Citations6Total Downloads320Last 12 Months5Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

A component theory of information retrieval using single content terms as component for queries and documents was reviewed and experimented with. The theory has the advantages of being able to (1) bootstrap itself, that is, define initial term weights naturally based on the fact that items are self relevent; (2) make use of within-item term frequencies; (3) account for query-focused and document-focused indexing and retrieval strategies cooperatively; and (4) allow for component-specific feedback if such information is available. Retrieval results with four collections support the effectiveness of all the first three aspects, except for predictive retrieval. At the initial indexing stage, the retrieval theory performed much more consistantly across collections than croft's model and provided results comparable to Salton's tf*idf approach. An inverse collection term frequency (ICTF) formula was also tested that performed much better than the inverse document frequency (IDF). With full feedback retrospective retrieval, the component theory performed substantially better than Croft's, because of the highly specific nature of document-focused feedback. Repetitive retireval results with partial relevance feedback mirrored those for the retrospective. However, for the important case of predictive retrieval using residual ranking, results were not unequivocal.

This paper demonstrates how a neural network may be constructed, together with learning algorithms and modes of operation, that will provide retrieval effectiveness similar to that of the probabilistic indexing and retrieval model based on single terms as document components.

An optimal query has been defined as one which will recover all the known relevant documents of a query in their best probability of relevance ranking. We have slightly modified the definition so that it also allows one to trace its evolution from the original to the optimal via the various feedback stages. Such a query can be constructed by modifying the original query with terms from the known relevant documents. It is pointed out that such a term addition strategy differs materially from other approaches that add terms based on term association with all query terms, and calculated from the whole document collection. The effect of viewing a document as constituted of components, and hence affecting the weighting and retreival results of of the optimal query, is also discussed.

A theory of indexing is presented and is based on viewing a document as constituted of components. A component may be chosen as any run of text unit that can be: (a) judged as to its relevancy property; and (b) considered as independent within the document. By looking at the constituent components of a document in relation to the universe of all components from the collection, we have been able to apply Bayes' decision theory to derive the index term representation for the document, as well as attaching an initial probabilistic weight for each term based on a Principle of Document Self-Recovery. It turns out that different choices of document components, such as a word or a whole abstract, can lead to different term weighting schemes that have been introduced before and are based on probability considerations; specifically, Edmundson and Wyllys' term significance formula, Sparck Jones' inverse document frequency, and later modified by Croft and Harper into the 'combination match' formula. Thus, a unified interpretation of various probabilistic term weighting schemes appears possible.

I NTRODU(3T I ON Most people will agree that the ultlmate goal of an Information retrleval (IR) system wouIcl be to Incorporate as much artificialIntelligence techniques as possible so that the system would be able to reply to queries with specific answers (so called fact retrieval).Current capabilities for large scale general purpose retrieval however Is still at the stage of document retrieval level, namely retrieving documents that most probably contain answers to a query.Thls may not necessarily be a transient stage at all, as has been pointed out that document retrieval capability may serve as a pre-processlng stage to the more detailed fact retrieval, even If the problems associated with the latter are assumed solved.In an automatic document indexing and retrieval environment, one general ly processes the natural language sentences from a query or from the title and abstract of a document into Isolated word stems (after removlng stop words), which are then used as representation of the query or document.This Is the content term selection stage of the automatic indexing prucedure.The next stage may be regarded as making the eviaence carried by each term more specific vla various term-weighting schemes.Since the words of the original sentences or phrases have been isolated, they lose their meaning In context.They become indicators or symbols of content rather than carriers of content, which is actually what indexing is about.As long as we are using these terms as Indicators, then there are other Indicators within a document that one can use.It Is well-known for scientific documents that their citing or cited documents are highly specific Indicators of the source document content.In particular, when two source documents share a number of identical references or citations, the probabil Ity of them deal lng with the same topical content would be high.When cited documents are used, this sharing Is called bibl Iographic coupl lng EKESS63]; when citing documents are used it is called co-cltatlon ESIVlAL73].In these methods it is Inherently assumed that the cited

Abstract A new model of viewing a document based on the citingcited relationship between documents is introduced. Using Bayes' decision theory, it is shown how a source document may be indexed and weighted by its set of relevant cited or citing document features, corresponding to a one pass relevance feedback Model 1 (probabilistic indexing) or Model 2 (probabilistic retrieval) system of [8]. Once every document in a collection has been so indexed, various forms of similarity measures based on probability of topical relevance between documents are derivable, including asymmetric, symmetric, and the relationship with Model 3 of [8]. Applications to retrieval and document clustering are also discussed.

Abstract A study was undertaken to classify mechanically a document collection using the free‐language words in the titles and abstracts of a corpus of 261 physics research papers. Using a clustering algorithm, results were obtained which closely duplicated the clusters obtained by previous experiments with citations. A brief comparison is made with a traditional manual classification system. It is shown that the mechanical procedure is capable of achieving simultaneous average relevance and recall figures above 80%.