Sunday, January 26, 2014

Fuzzy Logic







For many months one of the mathematics professors has been trying to find statistics on the numbers of community college students diagnosed as needing math remediation actually finished the recommended courses.  He came to me for help.  I searched the academic databases, Statistical Abstracts of the United States[1], and the U.S. Department of Education web site (http://www.ed.gov/) without finding any help for the professor.

This was a fuzzy search because we didn’t know the author, title, subject or even the best key words to search.  I tried searching terms such as “community college developmental math” and got more than 10,000 hits.  None of which were helpful.  Limiting the search to academic journals gave 3 hits, but none were even close to what was needed.  After many searches spread out over four days, I gave up; I had no more ideas.  The professor turned to his colleagues and one gave him a list of three articles that I was able to find for the professor.[2]

How does one find something when they are not sure of what they are looking for?  Our information seeking training does not offer great guidance. There is an express, “you can’t nail jelly to a tree” used for problems that are impossible solve.  What is interesting is source for this expression is Theodore Roosevelt, but many people think it is a myth.

May 23, 1912 Theodore Roosevelt had a long campaign day in Northern New Jersey.  At about 10:30 pm he got up to the podium at Dickenson High School in Jersey City, New Jersey to give a 30 minute speech.  He is a transcription of part of that speech.

When I became President I found the negotiations for the proceeding with great decorum as they had proceeded for years The Spaniards had discovered the Isthmus four and they had at once said that it would be very nice to have across it and there had been four centuries of conversation and I thought it was just as well that the conversation into action I did my level best to get Colombia to agreement. We were more than just we were more than generous Colombia. Finally I had lo make up my mind that to hold up Uncle Sam. I didn’t intend that Uncle Sam up and Colombia intended to blackmail a French company then have had France on the Isthmus. I was finally forced that to endeavor to negotiate with Colombia was about to nail currant jelly to the wall. You can t do it. It isn’t of the nail it's the fault of the jelly. [3]

I found this speech using several kinds of Google searches until I found a book that Google had digitized.  The academic databases yielded no results.

The problem with Roosevelt was that he was using conventional logic when he coined the expression.  Computer programmers and librarians know how to find solutions using unconventional searches and fuzzy logic.  There are solutions to this problem. One could freeze the jelly, one could put the jelly in a container and nail the container to the tree, or one could hammer a big nail into the tree and spread the jelly it.  Several people have said that “nailing jelly to a tree,” was one of Roosevelt’s favorite expressions, but I can’t verify this.  

How does one do fuzzy searches?  If you went to a store and didn’t know exactly what you needed, you could search the aisles in hopes that the product you need would be obvious.  Many times I don’t know the name of what I want; I just know what it is supposed to do.  One could go the grocery wanting food for dinner, but not know what to buy.  This is a retailer’s dream because the merchandizing display, special deals, and layout are all there to convince the consumer to buy something they had no idea they needed before entering the store.

The world of information seeking is a bit harder; there are no stores merchandizing information. There are no special information “sales.” One could browse the library shelves in search of a book and hopefully find something appropriate, but this process may be a matter of serendipity or blind luck.  One could read journals or other periodicals  and hopefully remember enough to find the article the next time you needed something similar, but few people have that kind of memory.  Unfortunately I have no great answers on how to find what you need with a fuzzy search.  One consults with an expert and gets guidance toward the correct path.  Librarians, subject specialists and colleagues are good sources for this type of guidance, but this kind of search depends on if you can find the expert and the time you need him or her. 

From the cataloger perspective I have wondered how the catalog record can be made friendlier to fuzzy search logic.  If the title is not very indicative of the subject and the subject headings are not what a searcher would guess, the search is hard.  We sometimes add tables of contents that are searchable with keyword searching, but this has its limits because the authors and/or publishers create these chapter titles.

Sorry, I have no answers that will work all the time.  Fuzzy logic and successful fuzzy searching depend on prior knowledge and wisdom that only comes with experience or access to those with the experience you seek.


[1] I looked in the print version of The Statistical Abstract of the United States first to get an idea of what was available then went to the U.S. Census Bureau web site that has the most current information.  http://www.census.gov/compendia/statab/

[2] One of the articles is from Community College Research Center, Teachers College, Columbia University.  “Referral, Enrollment, and Completion in Developmental Sequences in Community Colleges” by Thomas Bailey, Dong Wook Jeong, Sung-Woo Cho was right on topic.  The prepublication paper may be downloaded from: http://ccrc.tc.columbia.edu/presentation/referral-enrollment-completion-3.html The findings were presented on October 14, 2010.

The author supplied keywords are: Developmental education, Community college.  Searching on these keywords would give too many hits to be useful for the kind of information the math professor wanted.

[3] This is quoted in the book:  Theodore Roosevelt : one day of his life : reconstructed from contemporaneous accounts of his political campaign of 1912 …  / by William H Richardson. Jersey City : Jersey City Printing Company,  1921 page 34.


1 comment:

Minnowonsay said...

Interesting post, thanks. I think this is a bit like the concept of heuristic searching: (I see this is defined on Wikipedia as "a technique for solving a problem when classic methods are too slow" - [yes, I love Wikipedia even though, or perhaps because, I am a librarian, by the way].

I was taught that heuristic searching was why browsing was valuable: that it referred to a situation when a person comes in searching for one thing but what they learn during the search clarifies the way in which they need to search in future (i.e. they might start searching for a slightly different thing.)

This is what humans are good at and getting a machine or system to do it is very difficult (but not necessarily impossible.) The solution re-defines the problem. it's how humans learn and progress.