Many of you have used the IBM Watson Question and Answer (QA) service that allows you to ask questions and get answers on Bluemix, and have asked how and when it will be possible to bring in your own data.
This summer we released four new services that will replace the Watson QA service and help you customize and embed question and answer capabilities into your application. The new services can be trained on your industry or application specific data, and do not require deep knowledge in machine learning and linguistic models. As a result, the Watson QA service tile will be removed from the Bluemix catalog on Friday, November 20, 2015, after which you cannot provision new instances of this service. However, existing provisioned instances will continue to be available until December 16, 2015.
The four replacement services recommended for different types of question and answer capabilities are:
Natural Language Classifier – allows you to Interpret and classify natural language with confidence.
Dialog – allows you to script a conversation and help walk a user through a process
Retrieve and Rank – allows information retrieval with a machine learning model
Document Conversion – takes documents and ‘chunks them up” into smaller answer units to return as passages
When trying to build a question and answer system there are two different starting points which ultimately can help you build a comprehensive and robust system.
One approach is to start building out a system that is able to answer the most highly repeatable questions with defined answers. For this use case we recommend starting with a combination of Watson Dialog and Natural Language Classifier. You can merge these services together with speech services to create a voice based question and answer system. This will allow you to answer a large percentage of your user’s questions in natural language, and also enable you to learn from usage what other questions you need to train the system on. We have a few demo applications to get you started in the App Gallery: What’s in Theaters and Questions on the Natural Language Classifier are great starting points.
Another approach is to build a question and answer system based on documents such as FAQs or marketing materials. In this scenario you would use Document Conversion to transform your documents into Answer units, and use those as input to train the Retrieve and Rank service. As in the first use case, you will learn from usage how and what your users are asking, and use that as input to improve the system.
Also check out these recent blogs on applying the new “QA building blocks” services:
- Combining Natural Language Classifier and Dialog to create engaging applications
- Retrieve and Rank finds highly relevant content buried deep in complex information
We would love to hear from you! If you have any comments or questions, please reach out to us in the Watson Developer Forum, or leave a comment below.