Semantic components offer the possibility of allowing searchers to add additional specification to queries, which may allow search engines to more successfully match documents to the searcher's information need. This work will explore issues surrounding developing the sets of document classes and semantic components appropriate to a document collection, expressing information needs using document classes and semantic components, (manually) indexing documents using semantic components, and the effects of semantic component queries and indexing on searching. I have outlined our approach to investigating each of these subtopics and presented a few preliminary results.
In addition, I see areas of future work that will be beyond the scope of the dissertation, including automating semantic component indexing, investigating alternative versions of the model, and studying various ways of presenting and using semantic components in a searching interface.
I have shown that, using kind theory, we can use the same formalism to describe software components, reason about their composition, and generate verifiable “glue” code for their composition. These specifications come in the form of simple domain-independent annotations to typical program code, and such annotations are also used for documentation and testing purposes.
