Caroline Hagège

Normalization and Paraphrasing Using Symbolic Methods

We describe an ongoing work in information extraction which is seen as a text normalization task. The normalized representation can be used to detect paraphrases in texts. Normalization and paraphrase detection tasks are built on top of a robust analyzer for English and are exclusively achieved using symbolic methods. Both grammar development rules and information extraction rules are expressed within the same formalism and are developed in an integrated way. The experiment we describe in the paper is evaluated and presents encouraging results.

Intertwining Deep Syntactic Processing and Named Entity Detection

In this paper, we present a robust incremental architecture for natural language processing centered around syntactic analysis but allowing at the same time the description of specialized modules, like named entity recognition. We show that the flexibility of our approach allows us to intertwine general and specific processing, which has a mutual improvement effect on their respective results: for example, syntactic analysis clearly benefits from named entity recognition as a pre-processing step, but named entity recognition can also take advantage of deep syntactic information.

XRCE-T: XIP Temporal Module for TempEval campaign.

We present the system we used for the TempEval competition. This system relies on a deep syntactic analyzer that has been extended for the treatment of temporal expressions, thus making temporal processing a complement to a better general purpose text understanding system.

XTM: a robust temporal text processor

We present in this paper the work that has been developed at Xerox Research Centre Europe to build a robust temporal text processor. The aim of this processor is to extract events described in texts and to link them, when possible, to a temporal anchor. Another goal is to be able to establish temporal ordering between the events expressed in texts. One of the originalities of this work is that the temporal processor is coupled with a syntactic-semantic analyzer. The temporal module takes then advantage of syntactic and semantic information extracted from text and at the same time, syntactic and semantic processing benefits from the temporal processing performed. As a result, analysis and management of temporal information is combined with other kinds of syntactic and semantic information, making possible a more refined text understanding processor that takes into account the temporal dimension.

Linguistic and temporal processing for discovering hospital acquired infection from patient records

This paper describes the first steps of development of a rulebased system that automatically processes medical records in order to discover possible cases of hospital acquired infections (HAI). The system takes as input a set of patient records in electronic format and gives as output, for each document, information regarding HAI. In order to achieve this goal, a temporal processing together with a deep syntactic and semantic analysis of the patient records is performed. Medical knowledge used by the rules is derived from a set of documents that have been annotated by medical doctors. After a brief description of the context of this work, we present the general architecture of our document processing chain and explain how we perform our temporal and linguistic analysis. Finally, we report our preliminary results and we lay out the next steps of the project.

Using Morphological, Syntactical, and Statistical Information for Automatic Term Acquisition

Terminologies are useful in all areas that use specialized languages. The development of terminologies is a hard work, when manually done. It can be assisted with tools to ease and improve the achievement of such a work. In this article, we present ATA, an automatic terms extractor using both linguistic and statistical information.

Évaluation d’un outil d’aide á l’anonymisation des documents médicaux basé sur le traitement automatique du langage naturel

Anonymization of personal data is a legal requirement for their use as part of a research project. In the context of developing a tool for detecting hospital-acquired infections, 2000 medical documents were needed for the research project ALADIN. To help annotators to anonymize this corpus of documents, a tool for the anonymization has been developed, relying on Natural Language Processing techniques. The recall, precision and F-score of the automatic phase of the anonymizer were respectively 79.7, 85.2 and 82.4%. The gold- standard used for the evaluation was the manual anonymization of the documents. The performance of the automatic anonymization can still be improved but the tool is already a considerable help in this process in terms of saving time and in terms of quality of anonymization (including the accuracy of labeling anonymized terms and computation of time duration).

Portuguese Temporal Expressions Recognition: From TE Characterization to an Effective TER Module Implementation

Taking into account the temporal dimension conveyed in texts is a challenge to natural language processing. At the same time this task is of great importance for a wide range of natural language processing applications. The goal of this paper is twofold. First a characterization of Portuguese temporal expressions as they appear in texts is presented. This classification is intended to meet the requirements of high inter-agreement between annotators of temporal expressions. Second, relying on this characterization, an effective temporal expression annotation tool is described. Results from its evaluation are reported.

Linguistically Motivated Negation Processing: An Application for the Detection of Risk Indicators in Unstructured Discharge Summaries

The paper proposes a linguistically motivated approach to deal with negation in the context of information extraction. This approach is used in a practical application: the automatic detection of cases of hospital acquired infections (HAI) by processing unstructured medical discharge summaries. One of the important processing steps is the extraction of specific terms expressing risk indicators that can lead to the conclusion of HAI cases. This term extraction has to be very accurate and negation has to be taken into account in order to really understand if a string corresponding to a potential risk indicator is attested positively or negatively in the document. We propose a linguistically motivated approach for dealing with negation using both syntactic and semantic information. This approach is first described and then evaluated in the context of our application in the medical domain. The results of evaluation are also compared with other related approaches dealing with negation in medical texts.

Encoding and reusing linguistic information expressed by Linguistic Properties

This paper presents a way to express linguistic knowledge independently of any algorithmic machinery and of any particular grammatical formalism. This is performed through Linguistic Properties, that will be presented. First, the status of linguistic knowledge in grammars is discussed, then the Linguistic Properties are presented and two experiments are mentioned. They illustrate the reusability of the linguistic information enclosed in these Properties.