TTNS v1 Plan

This document describes the plan for a research project into state-of-the-art Technical Term Normalization System. The system must:

1. automatically discover of technical terms within research papers (tests will be restricted to abstracts).
2. automatically link those terms to the correct record in an ontology, if such a record exists.

The system can assume that:

1. one Ontology is provided.
2. a set of Documents (assume research paper abstracts)
3. a partial annotation of the technical terms in these documents, with some of them linked to zero or one node in the ontology.

For example ....

The approach that will be investigated are:

BASELINE A1) A term recognition algorithm that is based on a trained conditional random field approach to recognized technical terms. A2) A term normalization algorithm that is based on a Lesk algorithm that is optimized on the training data.

PROPOSAL 1) The loosening of A1) to present several options for candidates. 2) The creation of a graph an conversion into an edge classificatino problem.

---