Derek Hansen (2005)
Health FACTS: From Conversation to Archive and Back Again
I really liked the idea of using a distributed system to organize and catalog knowledge. For a topic such as health, I believe the benefit to the users will lower the threshold of the cost for participation. A general concern is the quality of information, much like the criticism that any such system (like Wikipedia) faces. However, for Wikipedia the cost of misinformation is much lower. When dealing with health, incorrect data seen by a person before it could be corrected could very negative consequences. In this regard, expert-based systems such as MCS and I-DIAG have an advantage. Of course, people should always consult and give authority to their doctor, but I fear this may not happen.
The ability to aggregate many keywords to a master topic is a great idea. However, I wonder if your solution will generate the optimal results. Perhaps using a mechanism such as
Everything2’s softlinks might also generate alternative relationships that will be bypassed using only keywords. Also, I wonder if your informing mechanism “21% of messages sent in the past year include this word” might misrepresent the usage. For example, what if a hot topic such as ADD is 21% of the traffic?
[I like the idea of using soft-links in addition to keywords. I don't think soft-links could be used in place of keywords since they require a pretty high level of activity on the site to work, but they could certainly help people make connections. As far as "fad keywords" go, I'm not too concerned since this statistic would only be used to provide a general idea of usage and using the past year will probably be close enough of a prediction of usage in the future. I could try different windows (e.g., % of messages in the past month or past 3 years) to see if that would be a better or worse predictor of future use.
DerekHansen]
Another (ubiquitous) problem is intention. A concept might be described across distributed terms within an email, such that no one keyword appropriately could be linked to the master topic. In this manner softlinks could have an advantage, as well as manual classification(s) according to a master index/directory. Furthermore, you hypothesize that the automatic insertion of links from keywords will help create a common vocabulary. Do you think that it also might cause clustering of keywords that are related to a master topic, but now the pointers become validated instead of consolidated? It would be interesting to see a method that would allow merging of related words. Additionally, do you think the convergence would still occur even with a large population of transient newcomers?
[Interesting idea. I'll have to think about what to do with clusters of keywords that consistently show up together. The level of convergence in language by various groups (e.g., newcomers) remains to be seen. On the one hand, newcomers will not be as familiar with the existing terminology since they've had less exposure. However, when questions are asked without using keywords, I expect that many of the answers (which are themselves more likely to be voted into the FACTS than the original question) will use the keywords, thus educating the newcomers. Additionally, many newcomers of the health lists I have come across actually read through the archives (in contrast to many other types of online communities), which means the keywords would be more likely to be used by newcomers than even some old-timers who don't read the archives regularly. In any event it is worth looking into once I get some data back.
DerekHansen]
Finally, I would like to see statistics on the likelihood of an individual to refactor a thread once nominated. For a person who is not highly active in a discussion or topic this might be too much work. This would especially be true if a person has received treatment for something minor and does not care anymore. While it increases the threshold for quality, you might also loose potentially useful threads. One method of quality control and content generation could be the creation of more experts in the system. Perhaps you might be able to collaborate with a medical school to create a student requirement to answer, consolidate, or edit information with later validation of the professor.
[I am also very interested to see who is willing to edit/post a message that has been nominated. I haven't seen any data on this yet. I agree that some people may consider it to be too much work, but am hoping that with a bit of positive peer pressure (e.g., listing the names of the people that nominated their message) at least most of them will edit/post their messages to the archive. The use of experts in the system is definately worth considering. The primary list that I have worked with in the past actually has several of its own experts (e.g., medical practitioners) that frequently post. Many lists are so specific I doubt that med students or even general practitioners would know as much as many of the members themselves. I suppose how big of a problem the quality issue is also depends upon the point of comparison. If one is comparing the quality of information posted to the FACTS to an existing email list (not using Health FACTS) then I would expect the quality to be better since only nominated posts are added and presumably they were nominated at least in part because they were of a higher quality. However, if one is comparing it to a medical encyclopedia I would expect it to be lower. The problem (from the community member's standpoint) is that medical encyclopedia's don't address many of the issues that members want to address, or they aren't comprehensive enough - and getting help from doctors is expensive.
DerekHansen]
(Comment by
AaronZinman)
[Thanks for the thoughtful comments. They've given me some new ideas on directions to take.
DerekHansen]
[Please insert comments in this section followed by your
WikiName]
There are no comments on this page. [Add comment]