njnubobo / online-lda

On-Line LDA: Adaptive Topic Models for Mining Text Streams with Applications to Topic Detection and Tracking.

On-Line LDA (OLDA)

• Rencently,I studyed LDA topic model and some variants of it,which has been widely used in textual analysis and data mining.

• This variants of LDA called On-Line LDA which focus on identifying emerging topcis of text streams and their changes.

• The original paper is as follows: L Alsumait, Barbar , Daniel , C Domeniconi.,On-Line LDA: Adaptive Topic Models for Mining Text Streams with Applications to Topic Detection and Tracking.

• Due to the limitation of the corpus,On-Line LDA model don't work well on task of my project.But it has inspired me a lot and I spend few days to implement the code.

• The original goal of my project aims to use dynamic topic model to analysis topic through Crowdsourced Time-Sync Comments(Bullet-Screen) which was segmented by timeslice.

• However,not all the code has been implemented(the KL-divergence of the evolutional matrix was not implemented).

Requirements

• numpy
• jieba
• uniout

Caveat

This is school project.If you have any question,let me know!

Author

WindWard [email protected]