[I forgot to Reply-to-All. Here's what I sent earlier.] This is a topic I've really been thinking about a ton lately in regards to Hackystat. I've been quite interested in both search topics and social network analysis connected to Hackystat. I remembered the "Expert Finder" you described here.
(I can't find the URL...it seems like it detailed a set of possible research topics for Hackystat) And I've followed some of the network analysis on political books done here. http://www.orgnet.com/divided.html I don't have time to say more now, but this topic seems really cool. The emergent structures of loosely coupled systems is also interesting to me. Jon Udell of InfoWorld has been talking about a few over the last year. Here are a couple links of his work. http://weblog.infoworld.com/udell/categories/infoworld/2004/08/30.html#a1064 http://weblog.infoworld.com/udell/categories/infoworld/2004/09/01.html#a1066 http://weblog.infoworld.com/udell/2004/09/20.html#a1079 http://weblog.infoworld.com/udell/2004/09/23.html#a1082 Could you combine these ideas with Hackystat into something that found experts for you? How about monitoring mailarchive.com, and looking at the mailing threads you were involved in? Combine that information with the java import statements in the code you touch, and many of the things you mentioned earlier. Can you create something that would find both local (Hackystat user) experts, and a search engine to locate a community expert? Experts on open source libraries, like Struts, can often be found outside the committer ring on the user lists. Are there other sources of information to locate these types experts who would have no direct code commit Hackystat info, but who could provide meaningful help? Anyway. It's a very interesting field, and I think it has a deep area of exploration available. Sorry for the short reply... --Tim On Thu, 23 Sep 2004 14:25:16 -1000, Philip Johnson <[EMAIL PROTECTED]> wrote: > Greetings, folks, > > I was ruminating on the idea of "social communities" today. Let's say that you want > to > work more efficiently. One of the best ways to work more efficiently is to figure out > what your current problems are, and then find an expert who you can ask to help you > out. > The depth and breadth of your social community will determine whether you know an > expert > to ask or not. > > These kinds of social communities are emergent properties of mailing lists and issue > management systems such as bugzilla. After monitoring such mailing lists and issue > systems for a while, you start to get a feel for who knows what. But this knowledge > is > incomplete and requires some investment on your part. > > I think Hackystat could be enhanced to support the generation and maintenance of such > social communities. One approach would be for sensors to monitor your activities > with the > goal of creating a profile of you that indicates things like: > > - what packages/APIs you appear to know about (based upon the import statements in > your > code and your code itself) > - what tools you appear to know about (based upon the tools you invoke, the command > line > monitoring, etc.) > - what errors/problems you appear to have encountered (based upon the error messages > that arise from Eclipse/Ant, etc.) > > You can use the Preferences page to decide what parts of your profile you want to > make > public. The public parts of this profile become available to other people on the > Hackystat server who are searching for answers to problems and can thus contact you > for > advice. > > To assess such a mechanism, a perfect environment is something like my 413 class, > where > students are constantly running into issues they don't understand and might want to > get > advice from other students who have encountered similar problems. Consider the > following > email I just received today: > > > Date: Thursday, September 23, 2004 12:45 AM -1000 > > Subject: ant question > > > > i keep trying to run 'ant dist', but i keep getting this problem: > > > > checkstyle: > > [checkstyle] C:\Program > Files\Eclipse\workspace\graphster-kusachi\src\edu\hawaii\graphster\FamilyTree.java:74:36: > Expected @param tag for 'args'. > > > > BUILD FAILED > > C:\Program Files\Eclipse\workspace\graphster-kusachi\build.xml:190: Got 1 errors > > > > ----------------------------- > > At 74:36 it says this: > > public static void main(String[] args) { > > > > I'm not quite sure how to fix this. Any suggestions? > > This student sent out this email at 12:45am. I was (fortunately) asleep at the > time, and > thus it mid-morning before I got around to answering this email. So, the student > waited > at least 10 hours for a response. > > Now, imagine that hackystat was building a representation of the "social community" > consisting of the students in this class. What if he could access the social > network at > 12:45am and find out which students (a) had run Checkstyle successfully, and/or (b) > had > run Checkstyle and had also gotten this same kind of error ("Expected @param tag")? > Now > combine Hackystat's social network with the fact that students already use instant > messaging, and this student could potentially start by finding out which students in > the > class had already encountered checkstyle errors, then look to see which of these were > online, then contact one of those students via IM to ask the question. The ten hour > request-response latency could potentially be reduced to a couple of minutes! > > There are lots of interesting challenges involved in developing such a social > network, > but I think this would be a fruitful research direction for someone because: > - Hackystat provides almost ideal infrastructure for the collection and analysis of > this kind of information. You could enhance the Eclipse and Ant sensors and get quite > sophisticated instrumentation support going quite quickly. > - After installing the sensors and setting the profile, no further user overhead is > incurred. Your profile is updated automatically based upon your behaviors and > experiences. > - A classroom setting involving programming is a really great environment for > evaluating this infrastructure, since it provides a ready pool of people confronting > common problems. And I conveniently teach such a class almost every semester. > - Such an approach would appear to find ready application to a lot of non-classroom > situations, even non-programming situations. > > Any takers? :-) > > Cheers, > Philip > >
