8:00 – 5:00 SR
- Still chasing down the Reqonciler issues.
- Need to put together a list of software that I have installed on my dev box and send to Lenny.
- Still working on the EXPLICIT entry. Added two manual items directly in the PHP and it’s working.
- Also see if I can get HTML parsed and displayed – done
- Based on that paper, I’m going to try keywords again. OK, I have keywords, but they’re really key phrases and as such, too unique? Here’s a screenshot:
- There are things that I think should correlate, like web-based, status, senior and visualization. There’s also ‘analysis’ and ‘analysys’ that should be matched by edit distance. So there are some options:
- Try a different NLP. Open Calais is free enough for what I’m doing, and does provide different parsing. FiveFilters has a PHP implementation that is 20 Euros, and actually seems to work best for what I’m looking for (items linked by keywords)?
- Or a naive kind of tagging that does some naive keyword extraction. These could be presented as potential tags to be checked by the users?