Ashika’s Status Update for April 4

This week I started integration with Abha for the text display. On my end, I created a display window (shown above) in Python that runs on a separate thread from the main program. This window is the same size as the pi and displays the sentences while KATbot speaks. In addition, I noticed that NLTK was not always outputting the right part of speech, despite reportings that it had an error rate less than 10%, so I played around with different types of default taggers. I found that most of them can tell when it is the wrong part of speech, but most output false negatives from time to time. Thus, I decided to use a combination of taggers, and the program accepts the word if any one of the taggers agrees with the target part of speech. I will formalize my investigations  in the final report. Apart from this, I started creating new templates. While doing so, I encountered some new minor bugs in the story program that I fixed. I chose one of the stories that has a long list of various inputs for our demo video.

I am still on track with the schedule. Next week, I will finish making templates and finish grammar correction. I will also work with Abha as she continues integration with the text display on her end, and improve the display to be a little more aesthetic. From now to the end of capstone, I will be mainly working on improving the system. I started using a children’s thesaurus rather than thesaurus.com for synonym generation, but I think I will do some more research into different thesauruses if time permits. In a week or two, I will work on unit testing. Because it is a little hard to get in touch with people at the moment, I may use a relatively small sample size of people to test story cohesion (maybe even use family members to grade them).