This is useful because I thought about something similar: clustering text, and using my auto-summarization library to extract key bits and then stich things together.
Now, I am not going to try. Lesson learned.
back
1 comments
This looks like a hobbyist effort built by somebody messing around with too big of a dataset. Anybody with even a smattering of NLP background could have predicted this quality of results. Computers are simply too stupid to assemble coherent narrative summaries on specific topics. They would require a true AI, and all of the major efforts that might build such a thing (Cyc, Semantic Web, etc.) have produced very little of value in the area.