A. Arampatzis, T. van der Weide, C. Koster, and P. van Bommel. An Evaluation of Linguisticallymotivated Indexing Schemes. In Proceedings of the 22nd BCS-IRSG Colloquium on IR Research, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....structures. Larger structures tend to capture more information about the original content of the sentence, while simpler structures are usually easier to index and search. Most semantically based Information Retrieval systems represent articles and references as unordered sets of sentences [12, 2], and do not encode information about how those sentences relate to each other. The knowledge representation should optimally satisfy the following properties: Canonical Representation: Two di erent sentences with the same basic meaning should have the same representation. Generalization: ....

....sentences that contain more matching structures are rated as more likely to be correct answers. This process is illustrated in Figure 1 1. 1. 2 Overview Although Semantic Information Retrieval techniques have always seemed very promising, traditional techniques have continually out performed them [27, 2]. In order to explore the limitations and the potential of semantic IR techniques, I designed and built SQUIRE, a Semantic Question Answering IR Engine. SQUIRE uses deep parsing techniques to convert sentences into sets of simple structures, representing the individual semantic relationships that ....

[Article contains additional citation context not shown here]

Avi Arampatzis, Th.P. van der Weide, C.H.A. Koster, and P. van Bommel. An evaluation of linguistically-motivated indexing schemes. Technical Report CSIR9927, University of Nijmegen, The Netherlands, December 1999.

....its object and subject. These pairs were then merged with indexing terms generated using other techniques (mostly statistical) Although the system only reported modest gains [33] in precision, it demonstrated the viability of applying linguistic analysis to information retrieval. Other authors [43, 2, 3] have also experimented with indexing word pairs that derive from head modi er relationships. The performance improvements have been neither negligible nor dramatic, but despite the lack of any signi cant breakthroughs, the authors armed the potential value of linguistically motivated indexing ....

Avi Arampatzis, Th.P. van der Weide, C.H.A. Koster, and P. van Bommel. An evaluation of linguistically-motivated indexing schemes. In Proceedings of BCSIRSG

No context found.

A. Arampatzis, Th. P. van der Weide, C.H.A. Koster, and P. van Bommel. An evaluation of linguistically-motivated indexing schemes. In Proceedings of BCS-IRSG 2000.

No context found.

A.T. Arampatzis, Th.P. van der Weide, C.H.A. Koster, and P. van Bommel. An Evaluation of Linguistically-motivated Indexing Schemes. In Proceedings of the 22nd BCS-IRSG Colloquium on IR Research, pages 34-45, Cambridge, England, April 2000.

....and the second is fuzzy matching which introduces a semantical similarity function between words into the retrieval function. Parts of this linguistically motivated indexing model are still under investigation, tuning, and evaluation. However, initial experiments have yielded promising results [2], suggesting that we are not far from nalising a model which should help to overcome the known and long survived problems of bag ofwords representations. 3.3 Matching The matching component, also called retrieval component, is responsible for a comparison between document contents and user ....

A. T. Arampatzis, Th. P. van der Weide, C. H. A. Koster, and P. van Bommel. An evaluation of linguistically-motivated indexing schemes. In Proceedings of the BCSIRSG '2000, 2000. To appear.

....accuracy. The removal of most indifferent terms is straightforward. The most common techniques use a stop list for removal of stop terms, and document frequency (DF) thresholding for sparse terms. Part of speech tagging has been also used for removal of common function words [Ruger, 1998, Arampatzis et al. 2000] However, after the removal of indifferent terms, a large number of non discriminating terms still remain. Automatic term selection methods can remove more of these terms according to training data statistics. Applying feature selection techniques to text classification tasks was found not to ....

....Koster, 1998] However, we found it inefficient to simulate a filtering situation (i.e. binary classification) with LCS, since we had to run the system for every topic individually. Therefore, we decided to run the final experiments reported in this paper with the same filtering system we used in [Arampatzis et al. 2000]. This system filters all topics in one pass over the collection. In this section we briefly describe the algorithms used, evaluation measures, the dataset, and pre processing. 5.1 Filtering System Our experimental system is based on the vector space model where documents and profiles are ....

Arampatzis, A. T., van der Weide, T. P., Koster, C. H. A., and van Bommel, P. (2000). An Evaluation of Linguistically-motivated Indexing Schemes. In Proceedings of the BCSIRSG '2000. To appear.

No context found.

A. Arampatzis, T. van der Weide, C. Koster, and P. van Bommel. An Evaluation of Linguisticallymotivated Indexing Schemes. In Proceedings of the 22nd BCS-IRSG Colloquium on IR Research, 2000.

No context found.

Arampatzis A., van der Weide Th. P., and Koster C.H.A., van Bommel P.: An Evaluation of Linguistically-motivated Indexing Schemes. Proceedings of the BCS-IRSG, 22nd Annual Colloquim on Information Retrieval Research. Cambridge. (2000).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC