Demystifying Topic Answers
Demystifying Topic Answers
Since the announcement of Microsoft Viva on February 4, 2021, I’ve come across a lot of questions and conversations surrounding Topics, specifically Topic Answers, and what a Topic Answer is…
In the past I’ve posted quite a bit on Answers and their relationship to Microsoft Search, whether File Answers, Acronyms Answers, or something else - see https://wbaer.net/2020/10/making-the-most-of-answers-in-microsoft-search/. To put simply, a Topic Answer, is just one of the many forms of “Answers” generated by Microsoft Search. To recap:
An Answer is a highly relevant and high confidence result that satisfies a user intent expressed as a query/question in search, presenting the most relevant information needed to get a job done and help users to faster task completion.
An Answer is a way to address user intent. When searching, the user typically types in characters and keywords to express an intent. Recognizing the keywords that are triggers for specific intents is important, but it is even more important that the content that is shown in search satisfies the user intent.
Answers are useful when you want to promote a search result to appear above ranked results. For example, for the query “sick leave”, you could specify a particular result, such as a link to a site that has a statement of company policy regarding time off work. You can think of Answers as being navigational aids to assist employees in getting directions to the information that matters most to help them keep productive and informed.
Topic Answers are a knowledge graph derivative that are intended to bring together people, content, and information. A Topic Answer has a name, definition, description and connection to people, content (such as file and sites), and conversations. The objective of the knowledge graph is to organize information, knowledge bases, and data from across apps and services within an organization. For example, Microsoft Search leverages the global perspective of the Microsoft Graph inclusive of world, organizational, and people knowledge.
The idea behind Microsoft Search and Microsoft Graph is that for search to provide great experiences, data is critical. This vast dataset is truly unique, powerful and secure.
The Microsoft Graph is a connected brain of world knowledge, organizational knowledge and individual knowledge The signals in the Graph, across Bing, Office 365, and Windows to name a few, enables ground-breaking possibilities for customers because they span the world around us, our daily work business applications, and user data. Tapping into data across these three areas helps us surface valuable insights and experiences to our customers across the Microsoft ecosystem.
And why is it a graph? It’s a graph because what’s really important is connecting these pieces of intelligence, so that these signals are not just individual points of information. The graph brings them together as something that we can draw patterns across. We can learn from one point of data to influence how we interpret another point of data.
Now coming back to Topics - a Topic Answer is not only one of the many Answers available to search, but one that aggregates many Answers, such as People, Files, and Acronyms.
So how are people connected?
A person is connected to topic when:
- A person is suggested as being related to a topic based on their contributions to content
- It is determined topic names based on content from files and contributors come from file activity
- The erson is added directly to a topic (admin curated)
- By confirming a person suggestion (AI suggested)
That said, a Topic and related entities, such as People, can be suggested, curated, or a combination of both. The former is where the knowledge plays a role.
The knowledge graph is organization-wide, but each topic can be visible to everyone or be scoped to a specific audience. Topics that are mined or otherwise AI suggested are initially available only to people who have permissions to the content, for example files, that support the topic. For example: at Contoso, a topic is “Mark 8”. This topic contains knowledge for everyone in the organization. However, a topic like “Increase Employee Education” is only available to people who have permissions to it.
The information in the knowledge graph is designed to be retrievable to help answer queries about what a project is about, what are the resources to learn more and who is related (and knows something about) the topic.
From a flow perspective, if the query expressed by the user matches a Topic or uses language that triggers asking for what a Topic is, the Topic Answer is diplayed. Herein lies the relationship to Answers… If the query previously triggered an Acronym Answer that can now be matched to a Topic Answer, the Topic Answer is displayed as opposed to the Acronym Answer.
To learn more about Answers see also https://wbaer.net/2020/10/making-the-most-of-answers-in-microsoft-search/.
Interested in learning more about Answers or have a specific question? Let me know in the comments below.