Expertise feature

The expertise feature has two major kinds of activities: administration and end-user discovery. The administrative actions configure the rules for the system and allow top-level reporting.

The discovery behaviors represent users looking for strong connections between other users and areas of expertise, as represented by single- or multiple-word terms (these single- or multiple-word terms are referred to in the remainder of this document as “tags”). Upon discovering a connection between a user and an area of expertise, users typically have options for communiticating with the person on that topic.

Function and configurability of the expertise engine

Expertise score calculation

A user's expertise on a topic is represented as a score that user has for a particular tag. Things that can add to that score include profile properties, actions taken by the user, and actions taken by others on the user’s content.

Administrators can determine which specific properties and actions affect the score, and by how much.

The actions that are scored can even occur outside of SharePoint. Skills management, certification, knowledge repository, and project management are examples of possible systems that could contribute external scores.

It is possible to drive expertise entirely from external inputs. To include an action that occurs outside of SharePoint, an administrator must set up an external input for that action.

The external system must then indicate to Spotlight when an incidence of that action has occurred.

Profile properties that contribute to the expertise score

By default, three profile properties contribute to the expertise score:

  • Ask Me About
  • Skills
  • Interests

Administrators can add other profile properties to this list, so long as those properties contain lists of text strings as their data.

As each profile property can only make one contribution to any particular user-tag connection score (or expertise score), these are weighted by default higher than actions, which can produce multiple contributions to a particular score.

Actions that contribute to the expertise score

There are two types of action:

  • active - Active actions are ones the user performs herself. These consist of
    • Microblogging (creating a microblog post);
    • Questions (asking a question)
    • Answer (answering a question)
    • Comment
    • Tag SharePoint content
  • passive - Passive actions are actions others take on the user’s content. These consist of:
    • Marked Answer (someone marked your question response as an answer),
    • Receive Like
    • Receive Comment
    • Receive tag on your SharePoint content
    • Your SharePoint content was rated

For each of these, “Weight” indicates the amount by which a user’s expertise score with a tag will be increased when that tag is involved with that action for that user.

For example, if a user answers a question that has the hashtag #competition in it, the user will get points added to his score for the tag “competition”. The amount of points added is equal to the weight for the “Answer” action (going by the picture below, this would be 20 points). The user will also get “Answer” action points on any hashtag the user includes in his own answer. They do not, however, get points against hashtags in other answers to the same question.

Weight changes apply retroactively

When Spotlight is first installed, the expertise engine begins functioning right away, counting numbers of occurrences of user actions from of a pre-determined list of action types.

These counts are converted into user- tag connection scores (or expertise scores), using default weights on these action types.

Occurrences of a tag within certain of a user’s profile properties also add to the expertise score, based on default weights for these properties.

The administrator can change these weights later, and have these new weights apply retroactively to events and properties that have already been counted by the expertise engine.

The administrator must be a farm administrator to be able to do this.