Example PhD

The Automatic Document Digestion

Supervisor: Professor N.J. Avis

Keywords: Semantic analysis, interactive documents, level of detail rendering

With today's current mass of information available at our fingertips, we need an automated way to assist us with digesting the salient points. If the user is faced with a 10,000 word document, how can then expect to absorb and/or find the important facts?

One solution is to go beyond linear reading of documents. What if the document is automatically processed down to a précis of 200 words? The user then clicks on areas of interest in the document, which the automatically expands the document in that area – so a word becomes a sentence; a sentence becomes a paragraph; and so on – so not a graphical 'Level of Detail' representation, rather more of a semantic summary 'Level of Digestion' representation. This approach affords the user to expand areas of which they are uncertain, and to collapse areas which are not of interest. This can be carried out automatically through learning the areas a user is interested in, or from 'weighting' certain keywords.

A custom version of a document can now be produced automatically for each user; the document becomes a 'living' entity – its presentation varies depending on the user, reflecting their current concerns/interests...

Key Skills/Background: Open to Computing Graduates and Postgraduates

Contact: Professor N.J. Avis to discuss this research topic.