I've been thinking about library systems and the information-use data they gather. Public libraries are very serious about keeping patron's information-usage data very private. In some cases, they even fight to protect people from prying government eyes. But should the case of privacy be entirely the same with information-use data in corporate libraries? Clearly, we want to keep some data private, but there is a lot of information-use and customization data that can be used to help people make connections to each other. As we all have learned with email, the issue of privacy is clearly not the same within the corporate firewall. In fact, there are enterprise software vendors built around the work of analyzing and extracting relationships in corporate email. However, with regard to digital libraries, I believe we still want to draw clear lines around information-use attributes that are appropriate for using in social networking and those that are not.

You may recall the entity relationship diagram I used in the CiL presentation to describe the attributes of business library users and their relationships to such things as topic interests, business units they work for, geographic area they work in. Some of these attributes are clearly extractable from sources such as HR databases. The library system web site/portal is also a potential source of information about people and their relationships to information. But we have to consider what types of information are easily extractable and appropriate.

The diagram below shows some these types of relationships that can be extracted from our HR database and library system. The Green boxes are those that I feel are appropriate for extracting now. With the exception of Documents, all of these can be extracted from our HR database. Documents may be extractable by our technical documents database and possibly from different document management systems residing on the Intranet. The Gray boxes represent descriptions of relationships people will have to indicate in the system for it to be public. This is the typical "profile" data people enter on social networking systems. The Orange boxes represent the more ambiguous or possibly appropriate areas and are the focus of my present concern.

Discussion group and usenet postings are also an area that I have concern about being inappropriate. Clearly we can data mine the discussion archives, but we do have some discussion areas, e.g. a rumor usenet group, where perhaps people want to be able to be fairly hidden. It's easy enough for someone to say that this is entirely appropriate because these are open forums.

The Subjects box refers to topic tracking interests. In our system, there are are a series of standard topic tracking pages that map to the technologies and topic areas associated with our business units. For example, one branch of our standard topic tracking area includes these Access pages:

Access/edge/core
* Access technologies
* Network management/OSS
* Storage
* Switching
* Video

We provide standard topic tracking for 109 topics, custom topic tracking for any topic that can be expressed as a Boolean search, and company tracking. I believe the standard topic tracking data is a potential source for connecting people, but those other topic and company tracking areas are probably not appropriate. Custom topic tracking allows users to enter any type of topic tracking interest. These types of interests might be very personal or reflect intersts that users don't want to share. We should protect those. Company tracking interests are also inappropriate, I believe. It might not be wise to expose, for instance, the companies that people are looking at in corporate strategy and development. Individuals also might not want to expose what companies they're looking at for job searching purposes. So I tend to think this should be private as well. We can create a profile area where people indicate the comanies they work with.

I can think of no reason why standard topic tracking Subjects shouldn't be used for making connections between people. Rather than requiring people to enter the topics they're interested in using their profile, we're using the topic tracking preferences they've already specified while using the library site and using that information-use data the way people use "Interests" on social networking sites such as Friendster.

The goal in this scenario is to expose some of the patterns of information seeking behavior to connect people with shared interests in the hopes that this supports community creation, thereby supporting knowledge sharing by helping people find experts. This is only one aspect of the social networking piece, but is one that I don't see discussed very much. Clearly in a corporate enterprise, this should be considered for its social networking and knowledge sharing potential.