Par4Sem

SEMSCH – Semantic Methods for Computer-supported Writing Aids

Semantic Methods for Computer-supported Writing Aids

Semantic Methods for Computer-supported Writing Aids (SEMSCH) is a research project funded by German Research Foundation (DFG) for three years from June 2014. The project particularly focus on the development of semantic writing aid plug-ins for word processors by exploring data driven, unsupervised and interactive machine learning approaches.

Acquisition of Semantic Similarity

The acquisition and utilization of semantic similarity in a semantic writing aid application can be achieved through different sources such as from structured data sources (Ontologies, thesauri, dictionaries, and semantic word nets), semi-structured data sources (Wikipedia, encyclopedia entries) and from unstructured text corpora.

Semantic similarities from structured sources

There are large number of structured resources available that can be used to encode semantic similarities on different levels. Top-level Ontologies such as YAGO, DOLCE and SUMO which include taxonomic relations between lexical words are used to encode semantic interoperabilities. Semantic WordNets and thesauri are another structure sources commonly used to compute semantic similarities. The research questions in this regard is the adaptation of such a resource on the target application as well as explore a systematic combination of several such resources.