nathanielwroblewski / chatbotta

ChatbotTA

A chatbot teaching assistant to answer student questions via slack

Setup

To run locally, you'll need:

• node and npm
• A Slack API key & bot user

First, export your Slack API key:

$ export SLACK_API_TOKEN="your-slack-api-key-goes-here"

Install dependencies:

$ npm install

Then, run the script:

$ node slackbot.js

Chat with your bot via direct message in Slack. Ask it a question, and it will respond.

Setting up your own Watson instance

For setting up your own instance of the Watson Retrieve and Rank API, grab your credentials (username and password) from Bluemix.

First, you'll need to create the Redis cluster:

$ curl -X POST -u "{username}":"{password}" "https://gateway.watsonplatform.net/retrieve-and-rank/api/v1/solr_clusters" -d ""

You'll get a redis cluster id in return. You'll need that for the following commands.

Next, you'll need to configure the Redis cluster and set the schema.

$ curl -X POST -H "Content-Type: application/zip" -u "{username}":"{password}" "https://gateway.watsonplatform.net/retrieve-and-rank/api/v1/solr_clusters/{solr_cluster_id}/config/example_config" --data-binary @./solr_config.zip

Create a collection:

$ curl -X POST -u "{username}":"{password}" "https://gateway.watsonplatform.net/retrieve-and-rank/api/v1/solr_clusters/{solr_cluster_id}/solr/admin/collections" -d "action=CREATE&name=example_collection&collection.configName=example_config"

Then you'll need to parse the forum data into questions and answers. Unzip the forum dataset located at ./forum_data/forum_data.tar.gz

Use the parse_answers awk progfile to parse the data for answers into a new parsed_answers.json:

$ awk -f parse_answers ./forum_data/forum_node.tsv > ./forum_data/parsed_answers.json

Use the parse_questions awk progfile to parse the data for questions with which to train Watson into a new parsed_questions.csv:

$ awk -f parse_questions ./forum_data/forum_node.tsv > ./forum_data/parsed_questions.csv

Upload the answers to Watson:

$ curl -X POST -H "Content-Type: application/json" -u "{username}":"{password}" "https://gateway.watsonplatform.net/retrieve-and-rank/api/v1/solr_clusters/{solr_cluster_id}/solr/example_collection/update" --data-binary @./forum_data/parsed_answers.json

Train watson using the following script.

$ python ./train.py -u {username}:{password} -i ./forum_data/parsed_questions.csv -c {solr_cluster_id} -x example_collection -n "example_ranker"

Training will take time (+20min).

You can then lob questions against the watson endpoint:

$ curl https://{username}:{password}@gateway.watsonplatform.net/retrieve-and-rank/api/v1/solr_clusters/{solr_cluster_id}/solr/example_collection/select?&wt=json&fl=answer&q={your question for watson goes here}

Or you can chat it up with the bot directly: