What's the Difference Between Search and Expertise Analytics?
During discussions with customers about the advanced capabilities of IBM Software Services' Expertise solutions, I find myself describing the differences between simple searches, enterprise searches, and the expertise analytics that we utilize. This blog summarizes those thoughts and explains why expertise analytics provides much better results than keyword or attribute searches.
Let's build up a series of layers representing different data sets and what results can be made available when trying to find an expert.
Directory - a simple attribute search
The foundation layer is the directory; that simple repository used for authentication and authorization which provides basic information like name, email address, phone number and office location. It provides answers to questions like "What is Scott's phone number". When extended to include job functions and titles, it can be used in a limited sense for expertise location by surfacing results to a search for "Certified Project Manager".
User Profile - aggregating what we know about a person
The next layer up is the user profile; a container into which we can aggregate additional details about a person. Solutions like IBM Connections provide a rich container of structured and unstructured data, aggregating details in three different dimensions
What the organization declares about me - the IBM Connections Profile is initially built from the corporate directory, with all the attributes from the directory layer incorporated into the profile. It is then extended using data integration tools to aggregate additional data elements from enterprise systems like Human Resources, Corporate Training, Skills Certifications, and Patent Databases
What I declare about myself - the IBM Connections Profile contains data elements for me to fill in, including personal background and resume information. I use social tags to stamp my profile with expertise topics, declaring the topics in which I have skills and interests
What my peers declare about me - the IBM Connections Profile then allows others to use social tagging to validate my skill areas. Peers add tags to my profile as I interact with them. The more peers tagging me for a certain skill, the more it becomes an indicator of my expertise.
With this rich profile in place I have a strong foundation for a profile search in order to locate experts. I can search for keywords and be presented with the people that have a match on that topic based on profile attributes and tags. The rich social profile is so important that we've developed a solution to rapidly capture information into it. More on IBM Connections Touchpoint in a future blog.
Content Contributions - expertise inference through publication
Any content that is contributed by a person is also a strong indicator of their expertise. IBM Connections have many containers for this purpose in the form of Files, Blogs, Wikis, Forums and Activities. Looking for an expert based on a topic keyword at this layer is similar to an enterprise search. It is very powerful and returns a large results set based on different content types. By focussing on the authors of the content, we can screen the results down to the most likely people having expertise in the needed area, and pivot to their profile to find out how to contact them. However, enterprise search results typically return a large volume of results, causing the searcher to segregate and apply filters to sift down to the real experts.
Recommendations - an application of profile and content analytics
So content and enterprise searches produce a lot of results that might overwhelm a user, and in large systems it is impossible to keep track of new people and new content as it is created. Hence the need for the Recommendations component within IBM Connections. This is an application of profile and content analytics to serve up important people and publications based on your areas of expertise. In this scenario, potential experts are served up for you consideration based on the attributes and actions you share.
The power of IBM Connections searching...
All these layers combine to form the core of IBM Connections, using search and recommendations to locate experts. It is very powerful stuff and many customers are at this part of the journey. They may have just determined a corporate directory strategy, deployed IBM Connections Profiles and begun aggregating additional data sources onto the profile, and started to use some of the content elements to capture the contributions from experts.
These elements form a repository of explicit expertise: we have specific tags, published content, and certification tracking levels stamped on the profile. What else do we have as an implicit indicator of expertise, or ways to make the results set of potential experts better?
...but let's harness the social platform and do more
Since IBM Connections is a rich social platform and we use these additional elements to reach out to more experts and refine the results to present the best experts to meet the need.
Social Validation - increase the expertise ranking based on the reaction of peers. As a content item is Liked, Reshared, and Commented upon
Social Network Analysis - knowing who is connected to whom, refine the results to promote the experts that have the most interactions with me
Engagement Analytics - based on the levels of participation in the system, promote the experts that are the most active and engaged. They are more likely to provide help
Inferred Expertise - Expertise in one area implies possible expertise in related areas. Add inferred experts to the results.
Prospective Experts - Who else has matching characteristics to known experts? It is likely that they have expertise, even if they haven't contributed yet. Seed the results with these prospective experts to get them engaged and active
Workload Balancing - It's not fair to route all inquiries to a single set of experts. Take their workload into account and even out the requests.
By taking these additional factors into account, the results set of experts moves from "Who has the skill?" to a much richer answer of "Who is the best at helping on this topic at this time?"
These additional factors are what IBM Software Services is incorporating in our series of IBM Expertise Solutions. Content within IBM Connections, as well as external content in enterprise systems, is analyzed and indexed according to standard full text practices and the more advanced social factors. We can then incorporate the expertise analytics into a variety of use cases:
IBM Expertise Locator - quickly find the best expert within the with organization on a specific topic and connect with them
IBM Social Connections Social Q&A - post a question and allow the system to route it to the best team of experts to get answered
- ...And more to come
For More Information
To get started, we recommend customers schedule a free IBM Expertise workshop. Contact the IBM Software Services representative in your area to learn more.