At this point, our research does not wish to perform blind text mining; however, we may wish to provide some indication of the type of text content in which we are interested.
Could you be more specific? If you don't want to perform blind text mining, then don't do it ;-). If you are looking for a particular text content, collect keywords that would provide corresponding indication.
For instance, given some lengthy medical text (document), a medical professional may only be interested in the aspects of the text that pertain to some particular case that they currently work on. It would be useful to be able to specify a query composed of relevant keywords, and then have an automated system provide an informative summary, unknown patterns or synthesis of the text, but limited to the context provided by these keywords.
Then I would perform two steps. In the first, you compose a corpus that contains only relevant documents; in the second, you assess frequencies of occurring words to detect relevant words or relevant patterns.
I see.
An important question, then, is what unit of text should be used for this tasks ?
- Badges
- Science topic
- Similar topics
- Social Science
- Sociology