Detecting deception in transcripts

Attached is an academic paper outlining how to detect deception in conference calls using transcripts, machine learning and linguistics. I would like to replicate this research to test the technology and its applications myself. The output of this project must be an easy to use interface (can be MS Excel, another program or custom a made app) where I can upload a transcript, edit or add/delete word categories. The interface will allow me to then run the analysis over the transcript and provide statistical information about the likely deceptiveness of particular words, sentences, and paragraphs. The interface should be flexible enough that I can conduct further research using a broader range of transcripts and word categories. Please read through the academic paper and respond with your relevant experience (using examples) and timelines.