FAQ answer system based on RapidMiner

nik600nik600 Member Posts: 1 Contributor I
edited November 2018 in Help
Dear all

i've recently discovered this amazing tool, and i'm asking if i can use it to classify question in mails.

What i'd like to do is, starting from an FAQ database build a classifier that given an e-mail it tries to find a related FAQ question, i think that the simplest approach is to put some keywords weight and relations to each FAQ, is it possible?

How can i do that with RapidMiner?

And, is also possible to build some "auto-training" ?

IE: train the system reading all emails in a specific folder assigning them to a specific FAQ question

TIA

Answers

  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,531 Unicorn
    Hi TIA,

    it's possible do to such things. You could design a process that will weight keywords automatically to suggest the most appropriate answers. But you have to take in mind, that the community Edition demands, that your system will be published under the AGPL3.

    With RapidAnalytics you can even build an interactive web site, displaying the most appropriate answers to any questions and letting the user choose one that helped him most, to take this into account for a re training.

    Even if all this is possible (comparably easy), it is nothing one would make in one day, especially if you are not familiar with this tool. And since it isn't that easy, I cannot post you a full description of what you have to do, but we are offering consulting serviceses for that.

    Basically you have to convert texts into numerical representation like TFIDF. This is possible with the Textprocessing Extension. Then you have to search for a suitable distance measure to find the correct answers. Of course the whole process needs to be optimized and validated.

    Greetings,
    Sebastian
Sign In or Register to comment.