One question that we often hear is, “How is Q&A in Talla different than search?” While both are ways of finding information that you’re looking for, Q&A gives users a specific answer to their question, whereas search provides a list of possible places that your answer might be. Search makes you do the work of finding the answer by digging through the results. Ask yourself: Would you rather have the treasure in front of you or a handful of maps that might lead to it?
In case it’s not obvious, having the answer presented to you directly can save you a huge amount of time. You’ve likely already experienced the benefit of Q&A over plain search results. Google recently implemented this in the form of featured snippets on their search engine results page. Previously, if you typed in a phrase like, “How do I export a Slack channel’s history?” you would have seen a list of search results with links to pages that might have the answer to your question. To find the answer, it was still up to you to click on the links, read the content on the page, determine if the page answered your question. If it didn’t, you’d continue clicking down the list of results and repeating the process until you found what you were looking for. Needless to say, this can be very manual process and extremely time consuming. This is how traditional knowledge bases work; you search for information on a topic, get a listing of search results, and go down the list clicking on each article until you (hopefully) find what you’re looking for. And if you didn’t happen to use the right search terms, traditional knowledge bases aren’t smart enough to know that a paragraph about SOC 2 compliance is relevant to someone asking a question about data security standards.
With Google’s featured snippets, when you search for an answer on Google, you’ll often times be presented with a direct answer to your question at the top of the page. Google has started providing this type of summary by extracting relevant information from the page.
Talla works in much the same way. When you ask Talla a question, either on Slack, MS Teams, or through our standalone web app TallaChat, Talla will understand the question you’re asking, extract the answer from your knowledge base documents, and present you with an answer to your question.
Being able to provide an actual answer to a question rather than just a listing of search results requires sophisticated work behind the scenes. We tackle this challenge from both the user query side and through smart content storage in Talla. Talla’s content training feature helps transform your unstructured documents into knowledge that answers user questions. Talla makes your documents smart by adding a layer of annotations that identify key entities (e.g. people, organizations, products) and conceptual tags (e.g. Sales, IT, Product) in your content. Unlike traditional knowledge bases, Talla uses machine learning and NLP to do the hard work and suggests the annotations to content writer. The writer can also improve content by adding in their own keywords, selecting entity types, and viewing suggestions to improve usability. This annotation layer help Talla’s AI understand your organization semantically and deliver the right answer when users ask questions.
We have also developed custom NLU (natural language understanding) techniques to better understand user questions and deliver precise answers. Our custom NLU is able to distinguish questions from search queries, infer the user’s intent, and return specific answers. When a user asks a question, Talla uses contextual and semantic clues in the user’s question to intelligently find the right answer in its knowledge base.
We’re doing the heavy lifting of understanding your questions and the content you create to save you time and make your team more productive. With Talla, your customer-facing teams can get immediate answers to the questions they're asking, which is much more efficient than just a page of search results. If you’re interested in seeing Q&A in Talla for yourself, schedule a demo today.