"BlogPulse Key Phrases, Key People, BlogBites, and Top Links are mined daily from new entries in over 750,000 weblogs using machine learning algorithms and natural language processing techniques. BlogPulse mines for bursty phrases and person names instead of for the most popular ones. The most popular phrases and names change very slowly over time. The burstiest phrases and names are those whose frequency of occurrence has increased significantly over the past two weeks, often dramatically."
I recently presented at the Computers in Libraries 2004 conference in Washington D.C. You can download the presentation, "Supporting enterprise knowledge management with weblogs: A weblog services roadmap", below:
My talk proposed a roadmap for providing weblog-related information services and suggested approaches for dealing with the problem of making weblog output of use to the organization. The idea is that the library can position itself to support individuals and communities of practice that express the need to use grass-roots tools for knowledge capture and dissemination such as weblogs and wikis. I talked briefly about the benefits of using weblogs for individual knowledge creation as opposed to using larger KM solutions selected from the top down, and the implications for IT of an information ecology with a diverse set of people using different technologies for publishing data in a distributed manner all over the intranet. In the near term I suggested first steps towards supporting knowledge creation with RSS. I suggested methods for providing access to aggregated blog output as next steps. And as a far off goal, I discussed the integration of output from sources such as blogs with other enterprise information using social software and social network analysis.
I was concerned, at first, that my talk wouldn't reach everyone. It was my perception from the conference that while many librarians are savvy when it comes to weblogs and RSS, as many are still learning what the value of the RSS format is. Some presentations dealt directly with the problems related to blogging on the one hand or with using RSS for keeping up to date. If these talks were successfully reaching people at the ground level (and that was my perception), my talk in comparison was probably much higher than most people wanted to go.
I talked about far off ideas such as how blogs might relate to social software and after talking at length with a good number of individuals after the session, it was clear that some people were definitely thinking about or actually have librarians weblogging at present in their organization. I found particulary interesting the talk given by Terence Huwe about his organizations' experience blogging at the University of California's Institute of Industrial Relations. Largely, the people I spoke to were interested in the topic of making blog output findable, but would wait until others proved the best directions to take. It didn't seem that the issue of the growth of customer weblogs (blogs external to the library organization) was on people's radar. This may reflect the type of information services organizations people worked in more than anything, but clearly this is also an early stage in our assessment. Even corporate libraries for large organizations, like my own, probably have not found the need to think about making weblog output findable yet. Many others will never have large enough numbers of blogs appearing in the organization to justify these kinds of systems. The issue for me was mainly bring to the forefront the idea that library organizations have the information management expertise to help with some of the problems blog output presents, and that we need to be thinking at this level sooner or later if we operate in an ecology where this type of grass roots knowledge creation might occur -- for example, in large corporations, university and institutional settings, etc. -- and if we wish to have a hand in innovating this new and potentially important knowledge space.
Some feedback received so far on this topic
In a discussion on Lou Rosenfeld's Bloug, Alan Gutierez suggests that an intermediary step before employing social network analysis is to utilize information managers who might serve as filters or recommenders. The idea of using people as recommenders is suggested in the "Mediate approach", which utilizes humans for indexing and supplying additional metadata. He also discusses tools that use ranking algorithms to analyze the frequency of memes and point back to the originators of ideas. I'm interested in this type of analysis as well. All very interesting thoughts.
One person suggested that the presentation should have pointed out more reasons why weblogging is a good thing. He notes that they: take the pain of publishing away; work great for small teams; utilize story telling, a powerful method for disseminating information; and allow multiple ways to publish. All valid points, and I believe I address some of these to a degree. The purpose of the presentation, however, is to focus on information services aimed at making weblog output findable, so I only touched on the main reasons I believe them to be valuable. But these points should definitely be included in a discussion of why weblogs are valuable as tools for KM.
Mopsos provides some additional thoughts about how much Information Architecture is related to the knowledge creation/management process, how innovation is supported by decentralization and re-use supported by centralization. He summarizes bottom up KM in this line: Decentralizing for innovation, centralizing for reuse.
He hones in on exactly the point I make about a mediated approach, which is takes the advantages of the decentralized approach to creation and the centralized approach to aggregation and dissemination. Very nicely put. To extract a few of his points:
bq. Technology is moving fastly into the hands of the individuals. ... What I will need is a common language, grammar and document structure, and I will be ready to forego some of my personal freedom to benefit from the advantages of using corporate standards.
bq. Value is added to existing information through a process of aggregation, simplification and dissemination, which is exactly the concept underlying a web of blogs. In essence, the more decentralized, the richer information is; the more centalized, the more structured it is. Decentralizing for innovation, centralizing for reuse.
This is a very good article by Mark Pilgrim that briefly describes the history of RSS and explains the schema/structure of an RSS document.
Robin Good's listing of the best RSS/RDF submission and directory sites.
Yahoo! Buzz Index is now offering the feeds of top searches in RSS format:
The News RSS Feeds are great if you want to follow a particular category of news. For example, you might want to read the latest Sports (RSS) or Entertainment (RSS) news in your aggregator. But what if you'd like an RSS News feed generated just for you? One based on a word or phrase that you could supply?
List or NYPL location with fee wifi services.
G2B Group white paper discussing the types of corporate communications material that could be served as RSS.
Taking some inspiration from Lou's EIA roadmap, I offer my view of the milestones in in the development of enterprise weblog information services. This illustration will be used in my Computers in Libraries presentation. While it provides the basis for my discussion around weblog services, it also allows me an opportunity to further discuss my view of information systems in the wake of recent doubts about KM as a valuable commodity in enterprise.
This is a short roadmap and the milestones are somewhat large targets, i.e. not considering the small steps in between. In my presentation I touch on some of the smaller steps necessary to get to the milestones and some of the cultural and political aspects that affect success at reaching each goal. This is mostly reactive strategy. It is a way to prepare for growth of grass-roots created knowledge and to meet the growth with information systems that can be used for sense making. I must stress that it is NOT intended to suggest that you must urge knowledge creation in the enterprise. I don't think that needs to be the goal of the information services organization.
I should note that the "near term" milestone suggests taking one or both of two paths: 1) providing enterprise blogging software and/or 2) being able to work with a diverse set of grassroots created knowledge. The second path has to do with my mention of diverse ecology, and alludes to the creation of a system that can aggregate XML data. My position is that an enterprise information services organization needn't push individuals to capture knowledge in blogs, but should instead react to organic efforts to capture knowledge. I should also note that this is not encouraging tacit knowledge capture. While some bloggers may choose to blog tacit knowledge if they like, the idea of forcing this kind of recording is the cause of many failed KM efforts. A good deal of explicit knowledge capture contained in writings, email discussion, formal publications, etc. is fodder for knowledge mining.
Some market researchers have been attempting to describe the failure of KM in recent years while others attempt to describe how KM is evolving now to allow diversity in the information ecology. But these reports remain nebulous about how to do KM right -- how to better extract and represented knowledge in information systems. For me, the process of envisioning a narrow information services roadmap has helped organize in my mind the concepts for information systems that take the burden of knowledge creation and organization away from the individual knowledge worker. A lot of my vision is interested in watching the interests of users in information nuggets they publish, in topic tracking interests they record, etc. The design concept is concerned with turning patterns of use into describable entities. For instance, "Michael Angeles (person)" is interested in "Information Architecture (topic)". The idea is to clearly define those entities and create rules for the extraction of those entities from various forms of published matter. That is to say, to let the system parse and extract terms and be able to distinguish that "Michael Angeles" is a known person or that "Information Architecture" is a known topic. Those entities can then used for social networking -- to interconnect experts in the enterprise. I discuss this idea a bit in my previous blog entry.
Extracting bits of explicit knowledge and inferring and suggesting the relationships of people to concepts, concepts to concepts, and people to people should be the role of the knowledge-based information system. This is the far off goal.