Don't reinvent the wheel
Elimination of redundant business practices lies at the core of knowledge management in large enterprises, and Net Perceptions
“We model people. We want to understand what they are working on. Are they studying rubber washers in an R&D type environment? Are they developing a new technology? Are they part of a sales and marketing team? Secondly, we model what they’ve done from a knowledge-work standpoint so that it can be reused later by others. In the process, we connect people who are about to do a specific project with those who have done it before.”
Version 2.0 offers new analysis and reporting capabilities that help management understand the relationships between people and knowledge resources. Managers can identify "knowledge communities" that don't exist on corporate organization charts-- which resources are most valuable to those communities. A collection function allows people who gather information pertaining to them to share personal knowledge with others in the enterprise. And, individual users can gain an understanding of the relationships between themselves and the people and resources within the organization, encouraging them to connect them with the right people and resources.
Sang explains that 2.0 is best suited to the largest of enterprises because as the number of employees grows, so, too, does the amount of knowledge that can be shared. “When you have a large sample size of people who want to research the same general area, the information that 2.0 will recommend will be that which is found to be most useful to the largest number of people.”
If, for example, a competitor comes along, dozens of people will all go to that company’s website and look for the most valuable information. Over time, all 100 people at the company might have done research, but different people will have looked at the information from different viewpoints. Knowledge Management 2.0 allows access to pertinent information from other perspectives or to people who might not normally interact--and may work on different continents.
“Plus,” says product manager Jim McKinley, “we take it a step further. We’re not just looking at everything an individual is doing, capturing it and recommending from that data set, we’re able to incorporate feedback. So, if a sales person or marketing research analyst goes into this competitor’s website and finds valuable content, we can recommend that they develop these communities from that value statement. So we’re not just pushing to people everything that’s been looked at.”
A number of variables weigh into that recommendation process. The first is the context of what that person is reviewing. Explains Sang: “Unlike a search engine or a taxonomy-based system, which organizes purely by relevance and degrees of relevance, Knowledge Management 2.0 incorporates feedback given by a human being about the value of this document. How relevant is it? How informative is it?”
Users are able to rate the document from a value standpoint—thumbs up or down or on a scale of one to five. The software can also infer value from a behavioral aspect. If a person is searching and clicks through on a document for a certain reason, maybe the author, the date, abstract, it gets a score. If she saves it, the document gets a higher score. If she accesses it repeatedly, it gets rated even higher. Additionally, documents can be presented back to the user. The software asks for feedback and, in real time, computes which pieces of information are best for that individual at that time based on the task.
Kim says, “Our system actually creates a model that defines the relationships across people and information. The knowledge base constantly organizes itself. By capturing how people interact with the information, the software builds connections so employees can benefit from what others are doing.”
Because only three to five pieces of information about every user are initially needed to get a system up and running, speed of deployment is prompt. At a large financial services organization, a prototype using company data was implemented 10 calendar days after the first meeting with Net Perceptions