kristoff@NET.BIO.NET (David Kristofferson) (05/24/89)
> Formalizing the AI of biology -- > and theorizing about reasoning on biological principles -- well, > I wouldn't want to point fingers or anything, but at this stage I > would much rather work with a few dozen system experts than a few > dozen expert systems. More fun, too. You are not the first person to express these sentiments. I am sure that this debate on AI will continue to rage. Also to respond to Dan Davison's last message, I fully appreciate that the goal of the BIO-MATRIX project is not to reason from first principles, etc. My point was that in most presentations of the matrix concept that I have seen/heard to date, the monologue starts out by talking about how wonderful it is that physicists can reason from first principles, etc., etc. This tends to go on for more time than necessary. In my experience as soon as one starts talking about such things to a biological audience people start tuning you out as being completely impractical. Later in the matrix presentation, one gets to the point about reasoning by analogy, improving biological databases, etc., which are all points that biologists would agree with. My question is simply: How many practicing biologists are still listening by the time this point is brought up? Again I believe that the standard Matrix description should "drop the physics talk" as I said earlier and get to the point of the project more directly. > So what will it take to siphon mainstream biological literature > into useful computer form? Perhaps there is some way I can help. I have heard several people suggest that this should be a major focus at the NLM. Can someone from there comment on what is being done to this effect? This is an extremely important topic in my mind. At BIONET I have taken a *very* small step in starting to introduce journal editors to the use of electronic mail and bulletin boards and in then getting them to submit their journal's Tables of Contents in advance of hardcopy publication. Perhaps next we can get the abstracts on-line (although I keep being told that this entails more expense for them). Of course, this effort is NOT considered *RESEARCH* (everyone bow with face to the ground) but is considered to be more worthy of assignment to a corps of thousands of faceless data entry clerks. There have been some cynics out there who have been less than supportive of these efforts. If one tries to get grant money for this type of "service work" one gets told that grants are for RESEARCH. If one tries to get the funding agencies to put out contracts, one finds out that this takes years to organize! Meanwhile the biological community continues to limp along. Steve Jobs will probably not come to your rescue because biologists are still rather poor and it is unlikely that a private company is going to make the investment needed to perform tasks of this magnitude when (a) there is a high risk that the investment may never be recovered and (b) the typical academic biologist doesn't fully appreciate the costs involved having never worked in industry (that place of the "wailing and gnashing of teeth" as it was portrayed to me before I "defected" ... best decision of my life, by the way!) and would rather try to do a band-aid job at best by paying slave labor wages to grad students who will do the project during centrifuge runs. Given this situation ye all shall reap what is sown. Will the funding agencies and the community get together on "service topics" like this? How long, Oh Lord, how long ...??? -- Sincerely, Dave Kristofferson BIONET Resource Manager kristoff@net.bio.net or kristofferson@bionet-20.bio.net
kristoff@NET.BIO.NET (Dave Kristofferson) (05/26/89)
> We are not just > after a big bag of data, we want to explain how things work. At a > minimum, the Bio-Matrix is a tool to facilitate discovery of biological > principles. > ... > Some of the important issues are: > > How do we enter the last couple centuries of information? Chris, If I read these two statements correctly you do not want to go after a "big bag of data" which is what entering the last couple of centuries of information would produce. On the other hand, being selective in what was entered would presuppose some higher order knowledge about what was "correct" in biology. If the Matrix project hopes to discover new knowledge by drawing analogies between disparate databases one runs the chance here of throwing the baby out with the bathwater by being "selective" during the entry process. I still think the Matrix goals are stated in a fashion that makes them unachievable by mere mortals, although later in your message you acknowledge the need for looking at "subdomains" in biological knowledge which is indeed reasonable. . . No immediate solutions to the monumental problems deleted here 8-)! . > One place almost everyone agrees we should start is by integrating > existing databases. The ISIS project, which has integrated 6 or 7 > molecular biology database into a monolithic whole, is one such > approach to this problem. What are other approaches to integrating > multiple, heterogeneous databases? More details about ISIS might be of interest to the general readership. Dave Kristofferson