Abstract. We discuss four linguistic ontology-mapping techniques and evaluate them on real-life ontologies in the domain of food. Furthermore we propose a method to combine ontology-mapping techniques with high Precision and Recall to reduce the necessary amount of manual labor and computation. 1

This paper addresses the issue of Web document summarization. As textual content of Web documents is often scarce or irrelevant and existing summarization techniques are based on it, many Web pages and websites cannot be suitably summarized. We consider the context of a Web document by the textual content of all the documents linking to it. To summarize a target Web document, a context-based summarizer has to perform a preprocessing task, during which it will be decided which pieces of information in the source documents are relevant to the content of the target. Then a context-based summarizer faces two issues: first, the selected elements may partially deal with the topic of the target, second they may be related to the target and yet not contain any clues about the content of the target. In this

We report on an approach to representation and retrieval of information from large textual databases. Our approach is based on dynamic ontologies that are automatically constructed from textual data by a new method combining techniques from knowledge representation, natural language processing, and machine learning. The method learns concepts automatically from documents, and uses them to build domain-specific ontologies and to organize the information contained in the documents. The ontologies generated are dynamic in that they are constantly updated and expanded as new documents are added, requiring minimal supervision from domain experts. Information contained in the documents are efficiently retrieved based on concepts in the ontology, allowing for precision and completeness to be traded off. A prototype implementation has been very encouraging.

Abstract not found

The major goal in ITC-irst's participation at SENSEVAL-2 was to test the role of domain information in word sense disambiguation. The underlying working hypothesis is that domain labels, such as MEDICINE, ARCHITECTURE and SPORT provide a natural way to establish semantic relations among word senses, which can be profitably used during the disambiguation process. For each task in which we participated (i.e. English all words, English 'lexical sample' and Italian 'lexical sample') a different mix of knowledge based and statistical techniques were implemented. 1

The accuracy of part-of-speech (POS) tagging for unknown words is substantially lower than that for known words. Considering the high accuracy rate of up-to-date statistical POS taggers, unknown words account for a non-negligible portion of the errors. This paper describes POS prediction for unknown words using Support Vector Machines.

Finding a proper distribution of translation probabilities is one of the most important factors impacting the effectiveness of a crosslanguage information retrieval system. In this paper we present a new approach that computes translation probabilities for a given query by using only a bilingual dictionary and a monolingual corpus in the target language. The algorithm combines term association measures with an iterative machine learning approach based on expectation maximization. Our approach considers only pairs of translation candidates and is therefore less sensitive to datasparseness issues than approaches using higher n-grams. The learned translation probabilities are used as query term weights and integrated into a vector-space retrieval system. Results for English-German cross-lingual retrieval show substantial improvements over a baseline using dictionary lookup without term weighting.

A common focus of systems in Intelligent Computer-Assisted Language Learning (ICALL) is to provide immediate feedback to language learners working on exercises. Most of this research has focused on providing feedback on the form of the learner input. Foreign language practice and second language acquisition research, on the other hand, emphasizes the importance of exercises that require the learner to manipulate meaning. The ability of an ICALL system to diagnose and provide feedback on the meaning conveyed by a learner response depends on how well it can deal with the response variation allowed by an activity. We focus on short-answer reading comprehension questions which have a clearly defined target response but the learner may convey the meaning of the target in multiple ways. As empirical basis of our work, we collected an English as a Second Language (ESL) learner corpus of short-answer reading comprehension questions, for which two graders provided target answers and correctness judgments. On this basis, we developed a Content-Assessment Module (CAM), which performs shallow semantic analysis to diagnose meaning errors. It reaches an accuracy of 88 % for semantic error detection and 87 % on semantic error diagnosis on a held-out test data set. 1

For the 11th straight year, the Conference on Computational Natural Language Learning has been accompanied by a shared task whose purpose is to promote natural language processing applications and evaluate them in a standard setting. In 2009, the shared task was dedicated to the joint parsing of syntactic and semantic dependencies in multiple languages. This shared task combines the shared tasks of the previous five years under a unique dependency-based formalism similar to the 2008 task. In this paper, we define the shared task, describe how the data sets were created and show their quantitative properties, report the results and summarize the approaches of the participating systems.

Inference tasks in computational semantics have mostly been tackled by means of first-order theorem proving tools. While this is an important and welcome development, it has some inherent limitations. First, generating first-order logic representations of natural language documents is hampered by the lack of efficient and sufficiently robust NLP tools. Second, the computational costs of deploying first-order logic theorem proving tools in realworld situations may be prohibitive. And third, the strict yes/no decisions delivered by such tools are not always appropriate. In this paper we report on an approach to inference in semantics that works on very minimal representations which can easily be generated for arbitrary domains. Moreover, our approach is computationally efficient, and provides graded outcomes instead of strict yes/no decisions. Our approach is fully implemented, and a preliminary evaluation of the approach is discussed in the paper. 1