Some Observations on ASK

(11:49:35 CDT)

Some Observations on ASK

NOTE: All of those currently taking LS507, you may look away. Heh, I don't need any more eye-rolling or flack about taking the readings too seriously.

For the rest of you, let's start with some definitions, ok? These are a combination of accepted and personal definitions. For now, I will not specify which is which, but roughly 70% is my own personal way of referring to things, not necessarily the accepted (though not against the accepted, as far as I know). Questions have two values, strength and plainness. The strength of the question is a factor of its conciseness, accuracy, and precision. Much more the latter two than the former one. Is a question on topic, is exactly what the person is asking, and is it in such a way as to avoid some confusion? The plainness of a question is its inverse relationship to its esoteric quality, so to speak. A perfectly plain question might be "What is 2 plus 2?" in which there is a easy to grasp, accepted answer. By the level of plain=0.9, the question might be "What is the GDP of South Africa?", which requires a little more digging but still a somewhat concrete answers is available. At plain=0.5, you might have "What is the most influential religion in the South Western United States?" a question that has an answer that requires some interpretation but could be measured in some way or another. "What is the meaning of life?" could be said to be a plain=0.0. There is no good way to sum up that answer without speculation, argument, and room for improvment. The more plain a question, the more likely the question will be strong, but not necessarily. It is easier to have a strong question with high degree of plainness, but it is also possible to have a quite weak question that is plain. It is also possible, though not necessarily easy, to have a perfectly strong question with a very low plainness (let's call plain<0.5 "exotic"), but it becomes harder. Also, in this line, the answer to the question has the same two qualities. Is it an exact answer? Is it simple? Is it on topic? Also, is it one verifiable things or does it require some inpretation.

I have been reading about ASK—aka anomalous states of knowledge—which has been discussed for a time but came out to some degree in a 1982 article by N.J. Belkin, R.N. Oddy, and H.M. Brooks called "ASK for Information Retrieval" (the first part was in Journal of Documentation, volume 38 (1982), number 2, starting on page 61, the second part was in the next issue). In it, Belkin et al describe wanting to know something as an anomalous state of knowledge. As my professor, Dr. Jeff Weddle, points out; the acronym may in fact be a backronym. At any rate, the questor realizes that he or she is missing something. They attempt to correct this by seeking a solution to their state. Belkin was writing, to some degree, from the viewpoint of a reference librarian, meaning the solution to the state involves an intermediary who has to translate the question into an algorithm of information retrieval; but we can assume that using search engines, scanning through books, asking friends, and such are all anagolous acts. In every case, we are trying to take our question, or recogition that something is wrong in our knowledge, and correct it.

I have discussed the Principle of Least Effort before on this blog (most recently on "Is the Internet Making us Dumber"), where we often take easy answers over harder but more truthful ones, and it comes into play here, most definitely. How could we know better? All we know is that we have an ASK. Being an ASK, we do not know what the answer is, so we take answers that seem to fit and seem to fix the problem. We can train ourselves to know better, of course, but that's something of a lifelong ordeal.

At any rate, it is more important to look at the concept of "best match" answers. A search engine tries to do this for us. Google often uses as "how often is this referenced/link" algorithm. Friends might give us the most interesting version of the information. A librarian might try and give us the most satisfying. None of these are necessarily the strongest answers. Traditionally, one of the roadblocks to proper IR is the idea that for every question, there is a single, best answer. If you ask, "What is 2 plus 2?" you expect "4". If you ask "What is the most important fiction book in the development of science?" you still expect there to be a good, clearcut answer. As the question gets more esoteric, more exotic, less plain, and weaker, less strong, the ability to generate a plain and strong answer diminishes. To paraphrase Belkin (and Dr. Weddle): we do not know what we are asking about, so we do not know how to ask the questions about it, and it makes it harder for them to answer the questions about it. Or, to put it more in Belkin-speak, "The assumption that the expression of information need and document text are functionally equivalent...seems unliklely, except [when] the user is able to specify [the need] as a coherent or defined information structure" (p64). Meaning that we assume that our need and the shape of existing knowledge are the same, which is untrue unless there is some, I would assume artificial, attempt at mapping the two.

The whole point of this is to say, keep in mind when you are looking for the answers to something, the more indefinite the question, the harder it is to make sense of the answer. We will never be able to ask perfect questions about things we do not know about (as I told a patron on the phone yesterday, one of our big stumbling blocks to looking up information is that we do not know all the proper terms, the proper questions, to ask). We can learn to ask better questions, but it takes time. My suggestion, tackle questions in stages. Do not be afraid to start off asking dumb questions and getting plain, but somewhat weak, answers. Then, using those terms, those ideas, ask bigger questions. Analyze small victories to synthesize the formula for bigger ones.

Si Vales, Valeo

• See other entries for June 09
• See Journal Archives (sorted by year then month)