Natural Language Processing for Knowledge Technologies

Objectives | Research Activities | Publications | Links | People | TCC Home Page

Objectives

The general problem we address is the design and implementation of algorithms based on Natural Language Processing (NLP) techniques that can be successfully employed to improve the interoperability between different repositories of knowledge, such as file systems, web directories, marketplace catalogs, etc., where a lot of implicit linguistic knowledge is present.
The objective is to use NLP techniques and semantic information to analyze this implicit knowledge, so as to find relations and mappings between the different linguistic expressions used in the repositories.
As regards semantic information we use WordNet, which, like other linguistic ontologies, is increasingly used to address a variety of content-based tasks (e.g. conceptual indexing) and to associate meaning with data and documents found on the Internet (e.g. in the context of the Semantic Web).
In this effort, we address the following topics:

• Multilinguality, i.e. treating linguistic structures that contain words in different languages.
• Word Sense Disambiguation, i.e. selecting the right meaning of polysemic words according to the context in which they appear.
• Terminology/Multiwords, i.e. dealing with multiword linguistic expressions typically used in specific domains.