EnterpriseSearchCenter.com Home
  News   Features   White Papers   Research Reports   Web Events   Sourcebook   Conferences  
 
Search All ITI Sites
Featured Content: Social Work--Adding Social Network Analysis to Search
Posted Aug 30, 2006 Print Version     Page 1of 1
CLICK HERE to download complete PDF

  

The web today is about participation and participant-created content. The most effective web search tools take this participation into consideration in the process of delivering relevant results. A look at these techniques (and some of the problems with them) can lead to insights into exploring the relationship between social context and search results inside the firewall as well.

The first name in web search, Google, examines participant behavior to enhance its search results. In social network analysis terms, it measures your degree centrality—how connected you are—through its page-rank approach. In addition to your degree centrality, Google factors many things into its search algorithm; it also keeps its algorithm secret to slow down those who want to cheat in order to gain a higher place in search results. Degree centrality, one of the central factors in Google's algorithm, measures the reputation of a site through who links to the site and who the site links to. Naturally, Google looks at other factors such as how relevant your content is to the search term, but your social standing in the internet is a key determinate of your place in its search results.

One of the constant battles that Google has, along with all internet search engines, is staying ahead of the cheaters who want to position their sites artificially high in the search results. It is relatively easy to game the system and get an artificially high degree centrality measure because degree centrality is a very direct and simple social network analysis approach. You can set up automated ways to create links and you can link into highly rated sites to benefit from their standing. Google does not make its social network analysis visible because its mission is simply to provide keyword search results. However, it also might not want to do this because it would give away tips to the cheaters on how they operate.

One-two Punch
There are other hybrid approaches to consider that make more direct use of social network analysis by combining it with search in a visible way that can be leveraged by the searcher. For example, iQuest uses a form of social network to enhance its search results and content analysis to enhance its social network analysis results. It measures the between centrality—are you linked to between important websites?—to gauge the importance of connections that pass through your site versus the whole web. Thus, the keyword in a given search field does not have to be on your site; it can be on the other sites that connect to and through yours. iQuest correlates well with Google but it is harder to game because of the indirect social network measures.

Both tools return the appropriate sites for the search term with links to those sites. However, a big difference between the two occurs in the results displayed via iQuest, which also shows you the social network analysis.

iQuest differs from other social network analysis in other ways, because it contains a content analysis tool that looks inside the content of electronic communication. Most social network analysis tools simply look at the relationships between participants in this communication. A hybrid tool like iQuest shows you who is talking to whom, what they talk about, when they talk, and where those conversations take place.

Searching for Social Structure
Social network analysis has some interesting applications for enterprise search, though when you go behind the firewall, the nature of content changes. Most of the content in the internet is unstructured and made up of large amounts of narrative data. Within organizations, new data factors in, such as company records and email exchanges. Behind the firewall, the mix shifts and there is more structured data. For example, you now find project status reports and financial results in spreadsheets.

Because their focus is on unstructured data, internet search engines like Google look at structured data as if it were unstructured, treating things like email addresses or phone numbers simply as data, rather than taking into account what they actually are. In a perfect world, a search engine would "know" whether it is looking at unstructured or structured data, and then understand and make use of the structure, if present.

There are a number of enterprise search engines that can address structured and unstructured data, but most of these are not capable of looking at the social networks around this data. Just like the web, the world inside the firewall is made up of social relationships. Hybrid tools allow you to both find content and benefit from understanding the human interactions around content. You can look at the email communication within a project team to see if it is functional or dysfunctional. Are key team members being left out? Is someone blocking communication? Are all the right people included? And, of course, you can see what they are talking about.

When you combine social context awareness with search, you can see who is central to the conversations on specific topics occurring inside the firewall and who may be excluded. Who are the real contributors? Who is being discussed, and what is being said about them? What are the most-used words on a specific topic or specific person? The integration of social network analysis with search opens up new avenues of inquiry and can put greater context to enterprise search.

See the Connections
Here is a simple table that summarizes the differences between internet search engines such as Google, enterprise search tools, social network analysis (SNA) tools, and hybrid tools that combine social network analysis with search.

This power naturally opens up issues of privacy, and policy needs to be made to govern proper use. However, with the new internet we are living in a world of greater openness and interaction, as evidenced by the growing popularity of tools such as blogs and wikis, not to mention social networking tools. It is interesting that in the early days of Quickplace, the Lotus collaboration tool, users were upset when senior management and IT people wanted to make access to the Quickplace collaboration sites more open. Now, companies are increasingly using blogs and wikis as collaboration sites because they provide the very transparency that used to have people up in arms.

We are moving into a new world of amplified social interaction and increased transparency. Combining social network analysis with search inside and outside the firewall will help us make greater sense of this world through improved understanding of the human interactions that occur around content.


Unstructured Data Structured DataLinks to and FromRelationships BetweenExact Data Not in Target 
Internet Search

Enterprise Search

some

Hybrid Tools

SNA Tools

Click here to download your free PDF of this article.

Print Version   Page 1of 1


Problems with this site? Please contact the webmaster. | About ITI | Privacy Policy