Evaluating Enterprise Search Solutions Today - Q&A with Dr. John Lewis, Chief Knowledge Officer of SearchBlox
One of the main ways that AI supports the search administrator is in automatically generating document titles and metadata.
JW: Everyone is talking about AI-driven search now. How does SearchBlox use AI to make searches more intelligent?
JL: We think of AI as supporting both the end users of search and also the search administrators who may also be called content managers or knowledge managers in some organizations. One of the key ways that AI supports the search administrator is in automatically generating document titles and metadata. It's a classic problem in the search space that we call garbage in, garbage out, as far as document titles. People are searching for something with a search query, and it has to match something in a document. If you have a bunch of documents with no titles or very poor titles, they're just not going to come up. People will say, "Search isn't working," and there's very little that the search administrator's going to do unless they are willing to stop and work through hundreds or thousands of documents to try to do that manually. AI can automatically generate titles and metadata by looking through the documents and finding a good answer for what that should be so that those documents come up.
We use synonyms to find similar words and documents, for example, and we've had this for a long time, where someone might search for something like "change password," and the document is titled "update password." Things like that have been around for a while. Synonyms work if the match is close, but AI is needed when it's not even close. Using AI to generate titles is one way we support the search administrators.
JW: That's great. And how would you say that AI is used to support the end users of search?
JL: With traditional search, you type in some text, a list of search links comes back, and that's the end of the process. At SearchBlox, we have an AI search strategy that we call DirectIntent, which is working throughout the entire search journey. And it has four stages that are broken up. We haven't maybe really thought about it, but there's a stage called Before Typing, and then During Typing, and After Typing, and After Selecting—after we go through it. How can it help you before typing? Just by clicking in the search box, it's giving you a list of search suggestions, maybe based on popular searches, things that people are looking for, or promotional content and things like that. We think of it like widening the zoom lens. Maybe you thought you were going to start with the search term, but it gives you some options up front that may widen the view of what you thought maybe you should be looking at.
DirectIntent is SearchBlox’s AI-driven search strategy that enhances the entire search journey.
Then, during typing, you're still getting a list of suggestions, but it starts to narrow down. Of course, after typing, you get a list of related documents, but also the AI can bring up specific answers. With traditional search, it says, "Here's the list of documents," and the number-one link on the list is probably the most relevant, you click on it and then you look through that big document trying to find out why the search engine thought that this has the answer in it. But with AI, you can bring back the specific answer that it found in that document, as well as a list of links. Then, instead of just stopping there, a search can continue working by looking after you select something. Maybe you clicked a couple of links, or you went to the second page of results and clicked on something. That's all-important information because it's monitoring the entire search journey. With that, it can produce insightful reports about what you're really looking for when you change your mind while searching, or if you have to go through a couple of pages to find the answer you're looking for. That provides insightful reports that lead to continuous improvement. DirectIntent is our AI-driven search strategy that enhances the entire search journey.