Hoory Knowledge Base is a unique application designed to crawl the content of the website where the widget is embedded.
Basically, this information is used to create a database of knowledge and is used to send relevant information to the users’ requests when there are no suitable Trainings available.
KB in the Workspace
To create a unique database of knowledge for your assistant, simply let Hoory start the very first crawl!
Just click Start crawling and let Hoory handle the progress. Meanwhile, you can easily navigate through the workspace. You will be notified once the crawl is complete.
Note: The time of crawling can vary due to the volume of your website’s content.
After crawling the widget-embedded website, Hoory returns a list of crawled pages that build the workspace’s knowledge base. The relevant paragraphs from these pages will be sent as answers to the questions of your users.
Take into account that you can Hide from Index the pages that you don’t want the users to receive. If a page appears with a No Index badge, it means that the page has blocking mechanisms which prevent Hoory from crawling.
In the list of crawled pages, you can view the number of questions matched with one page, which shows how many times the page has been matched as relevant and sent to the users. You can also view the date and time of when this last happened in the Last matched column.
In the nested view of the list, you can also view the given subpages and select multiple pages to perform an action.
Also, there’s an option to preview the pages and the crawled HTML codes directly from the list.
Start a New Crawl
You can always ask Hoory to recrawl your updated website by clicking Start a New Crawl. This way, Hoory will rewrite the Knowledge Base by crawling the website from scratch. This process can be stopped at any time and it will return the list of pages successfully crawled up until that moment.
Once the KB is enabled, your assistant will send relevant pages to the users’ requests with a badge saying Retrieved from Index. In the page preview, Hoory will highlight the part of the page where it found the relevant content to the question. Your users can then open the page and see the corresponding paragraph.
Note that your assistant queries the trainings first, then switches to the Knowledge Base to find answers. So, when the widget users ask a question that has no trained answer, Hoory sends the already crawled pages that are relevant to the request.