Some things read this week, 12 -18 August 2007

Posted on by

Monday, 13 Aug

Wilson, Patrick. “Situational relevance.” Information Storage and Retrieval 9 (1973): 457-471.

Cited by Raber in The Problems of Information, ch. 9, en13, p. 186 for the whole article in support of:

While many people, for example, may share the same problem or at least the same kind of problem, it does not mean that the same information will be useful to each of them in exactly the same way. As a result, relevance must be regarded as individual and situational, depending on the user’s perceptions, concerns, preferences, current state of kowledge, and view of his or her situation (186).

I like a lot about this view of relevance, except for its reliance on question answering. Perhaps parts of this view can be used while expanding beyond the question answering, but probably not without throwing out the question answering logical basis. I am unsure whether the logical basis is supposed to be prior to, or whether it is, or could be, after-the-fact. I feel that it should be, primarily, after-the-fact, that is, the sort of post hoc conditionals that I was complaining about last week.

Froehlich, Thomas J. “Relevance Reconsidered—Towards an Agenda for the 21st Century: Introduction to Special Topic Issue on Relevance Research.” Journal of the American Society for Information Science 45, April 1994: 124-134.

This is the intro to a special issue on relevance research. This article and many of those comprising this issue are cited by Raber in ch. 9. This article in end notes 11, 14 and 22.

Sets a good stage for the articles in the special issue, and serves as a good summary itself. Worth a read by itself.

Priss, Uta. “Formal Concept Analysis in Information Science.” In Cronin, Blaise, ed. Annual Review of Information Science and Technology (ARIST) 40, 2006. Sorry, can’t give a full citation since I read a draft version and it’s 11:30 PM, but I verified it here.

I read this due to something my advisor pointed me at but I sure hope this isn’t what she meant. She’s on a very well deserved vacation so I’ll have to wait a few more days to find out. [As I suspected, it was not.]

I have no doubt that this technique can be useful but it just perpetuates many of the things that I am beginning to see as wrong with what we do in LIS. Just because something can be done easily in a computer is not a good reason to do it that way. And, honestly, the mathematicalization of language and concepts is just too much.

Formal concepts in FCA can be seen as a mathematical formalization of what has been called the “classical theory of concepts” in psychology/philosophy, which states that a concept is formally definable via its features (draft 11).

The advantage of formalizations, however, is that notions are defined with absolute precision within the formal realm and that they therefore may be implementable in software (draft 12).

Ugh!

Svenonius, Elaine. “Reference vs. Added Entries.” [link] Paper presented at Authority Control in the 21st Century: An Invitational Conference, Dublin, OH, March 31-April 1, 1996. [originally read 11 May 2007].

Directly suggested to me by Bryan Clark Campbell [Sorry, Bryan. Losing my mind.]. I just wish this paper didn’t end in the “middle.” Not sure how much is really missing, but it clearly ends abruptly.

Tuesday – Wednesday, 14 – 15 Aug

Raber, Douglas. The Problem of Information: An Introduction to Information Science. Lanham, Md: Scarecrow Press, 2003.

Ch. 10. on “Information as a Social Phenomenon.”

Wednesday, 15 Aug

Jin, Qiang. “Eliminating redundant entries in bibliographic records.” Library Collections, Acquisitions, & Technical Services 29 (2005): 412-424.

Also suggested by Bryan Campbell along the same lines as Svenonius, that is, pulling apart the function of references vs. added entries.

Thursday, 16 Aug

Ellis, David. “The Physical and Cognitive Paradigms in Information Retrieval Research.” Journal of Documentation 48 (1), March 1992: 45-64.

Cited by Raber, ch. 9 en15, regarding the physical and cognitive paradigms not exhausting the ways in which to think about information as a theoretical object.

I guess I should add that I read a little of David Bade’s The Theory and Practice of Bibliographic Failure, Or, Misinformation in the Information Society each day. I guess you could label it my “bus ride” book, although I do read it on a few other occasions. One shouldn’t rush through a book on errors, though, it seems to me. T’would be an error; would it not?

Friday – Saturday, 17 – 18 Aug

Frohmann, Bernd. “The Power of Images: A Discourse Analysis of the Cognitive Viewpoint.” Journal of Documentation 48 (4), December 1992: 365-386.

OMG! If only Raber could write like this. At least he cited it 4 times in ch. 10.

Lots of connection to Dr. Richard Stivers’ work and the things I did with him. Will have to go back and re-read a few things of his.

This is an incredible analysis of the cognitive viewpoint in LIS.

Highly recommended, but you really ought to read up a bit on the cognitive viewpoint first. Frohmann does outline it, of course, but in a fairly cursory way. I would not have been as impressed with the analysis if I hadn’t already had a good idea of what was being critiqued.

Saturday, 18 Aug

Robertson, S. E. “Between Aboutness and Meaning.” The Analysis of meaning : informatics 5 : proceedings of a conference held by the Aslib Informatics Group and the BCS Information Retrieval Specialist Group, 26-28 March 1979, The Queen’s College, Oxford. Maxine MacCafferty and Kathleen Gray, eds. London : Aslib, 1979: 202-205.

Cited by Raber, ch. 7, “Representation of Information,” en5, p. 133. “On one hand we can say that the purpose of information retrieval systems has little to do with answering questions, satisfying needs, or even resolving anomalous states of knowledge. Rather, its ultimate purpose is to retrieve texts that will help users of the system do these things.”

This is what this short paper claims, but I’m thinking this is a bit narrow. Perhaps it is because I am, in a sense, inside the system, but I often use our systems as a typical user and as a cataloger (another kind of user) to do just that. I frequently look up a surrogate of an item (on the web and in the OPAC) to answer a question, satisfy a need, and/or resolve an ASK. I have no real desire in the document itself sometimes, just in a specific piece of metadata about it. For instance, I may look up a record of an item so I can import it into Zotero. Or, I may need to know if we have a previous edition of an item so I can assign the same call no. Or, do we have another item on this topic by the same author so I begin with the same cutter. In these cases, I could care less about retrieving the item itself.

As we expand our concept of information retrieval systems beyond the idea of an OPAC and databases these sorts of examples should proliferate. How about it? Can anyone think of any other examples of using an IR system (typical library or web 2.0 or otherwise) in a more direct fashion? That is, not to retrieve the document or text but to answer the question directly, resolve an ASK, etc.

Looking something up in Wikipedia fails as one is retrieving a document there. I guess one could argue that the surrogate that I retrieve in the OPAC is also a document. Sure. In one sense, I agree. But I think that’s fundamentally different than a Wikipedia article, say. And I think if I let the above view off the hook so easily then we are unnecessarily restricting our vision of what an IR system can be and be used for.

Any thoughts?

7 thoughts on “Some things read this week, 12 -18 August 2007”

  1. “Just because something can be done easily in a computer is not a good reason to do it that way.”

    Do you have an example in mind, Mark?

  2. Jodi, I intend to address your question because my really bad shorthand there deserves to be fleshed out. Glad you pointed it out. ;)

    But I’ve been forcing myself to do résumés and cover letters since thee jobs close soon.

  3. Jodi,

    A lot of stuff compressed into those short comments. Not even sure I can, nor want to, do them proper justice as there are several complex issues entwined.

    As for any specific examples in mind … nope, not really. My concern is mostly with the attitude that when something is computerized it is done so in the easy manner, precisely because it is easy and not because it a proper way to do it.

    I am not saying that many things should not be computerized. Although some things should not be.

    I am also not saying that things that are easy to computerize should not be done.

    There are often social, historical and technological reasons why something is computerized using an easy method. It may be only way to do so with the current resources, be it because of an understanding of programming principles, program language availability/capability, cost of (many factors), computational power, storage power, lack of bandwidth, etc. Many of these factors–or even one–can lead to something being implemented in an, ultimately, less-than-satisfactory way.

    There are many places I could go with this and many sources I could cite (if I could only remember them or had the time to look around and re-read some things), but I’d like to just offer a taste of part of my criticism and a reason for why some of the things of most importance to us in LIS, and in language in general, are and ought to be extremely hard computationally. This extended quote comes from David Bade’s The Theory and Practice of Bibliographic Failure, or Misinformation in the Information Society, and is from Section I.B.7.c “Metaphor and variation: freedom to err,” pp. 68-69:

    This brings us back to the comment by Hollnagel noted above, namely that error is a meaningless concept because it does not have a single meaning. Hollnagel expresses the desire of the mathematician and technologist; perhaps also for many scientists who believe or assume that all language must have the character of number. On this matter de Koninck and Duclos have offered some comments that are especially relevant to the construction of library catalogs:

    The richer a language is, the more polysemy and nuance, the better it serves thinking. The cybernetic revolution, by universalizing language as 0 and 1, has created a Newspeak of a different type than that of 1984, but one no less impoverished and totalitarian. 144

    The binary ideology transforms man into software so that he becomes a factor of certainty, of diagnostic and production. By making him mechanical alongside the machine, not only in his actions as in Taylorism, but in his spirit, by this Newspeak, … he is civilized to no longer pose problems of reliability for sociotechnical integration. 145

    Of course the evidence is abundant that all language is metaphor and variation, that there are multiple norms depending on political, geographical, regional, religious, ethnic, economic, racial, psychological and historical factors including gang membership, sexual intentions, oral surgery, audience, stylistic intentions and especially the place, time, manner and reasons for learning and using a language in the first place. Translating the language of science into correct “usage” is a pretentious absurdity and demonstrates the philosophers’ desire to legislate and control rather than the desire to understand just as much as the insistence on single invariant meanings among scientists reflects their obsession with controlling rather than knowing nature. 146

    144 De Koninck (2000), p. 64.
    145 Duclos (1999), p. 25.
    146 This pretension may be contrasted with the practice of translating the language of a particular text into the language of a particular set of subject terms for purposes of human or mechanical indexing and retrieval by subject: this task is beset with all the problems of misunderstanding, metaphor, conflicting usage and linguistic variation involved in ordinary human speech but it never tosses out the item to be indexed because it does not fit easily into the indexing language. Librarians may not be philosophers but the work they do is certainly more useful.

    Now, there is a lot of previous critique included in this excerpt, but it hopefully points at some possible (and imho, accurate) criticisms of these views.

    The bigotry involved in a statement such as the following is incomprehensible to me:

    The advantage of formalizations, however, is that notions are defined with absolute precision within the formal realm and that they therefore may be implementable in software (Priss, draft 12).

    I may not be a computer scientist, a linguist, or even a “real” philosopher, but I’ve been attempting to use at least one natural language for long enough, and I have a bit of experience with formal ontologies, Topic Maps, etc. to realize that the attempt to, on a wide-scale, define concepts with absolute precision is a complete non-starter.

    In extremely limited domains this is possible, to an extent. But in our field, where are the domains of use that are this limited?

  4. Re: “Just because something can be done easily in a computer”

    There is actually a very large literature on the kinds of problems which follow from acting upon the assumption that if it is possible to automate/computerize something it should be. Most of those problems appear to arise from understanding a particular action/process in isolation and not foreseeing how the automation will effect the whole environment, everything from carpal tunnel syndrome to market crashes following automatic trading. In a small task, automation may be far more expensive than the performance of a skiller laborer (e.g. subject analysis of any particular book is a small task, yet one which requires a world-view adequate to the task). Dörner’s research demonstrated dramatically that not seeing the connections between actions here and now with the then and there was a primary factor in the production of management disasters. The most general critique that has been made about automation in general is that it is only 100% successful when everything that needs to be known IS known and IN ADVANCE. What cannot be predicted with 100% accuracy will require the supervision and sometimes intervention of a human operator, but that human operator is no longer involved in any processes except those which the computer cannot properly execute. The operator is thus left with a completely arbitrary collection of tasks which could be fairly described as correcting the computer’s mistakes and filling in its lack of knowledge in response to real time events. The operator is supposed to trust the computer to do the right thing because that was the whole reason for automating in the first place, yet his only real task is to recognize a problem, interpret it and correct it.

    In addition to Dietrich Dörner, three seminal and influential essays are recommended by IDQ:

    L. Bainbridge, “Ironies of automation” in Automatica 1983 v.19 nr.6 p.775-779

    Parasuraman and Riley [Victor, not Jenn], Humans and automation: use, misuse, disuse, abuse” Human factors 1997 v.39 nr.2 p.230-253

    Poyet, Christine. “L’homme, agent de fiabilité dans les systèmes automatizés” in J. Leplat and G. de Terssac, Les facteurs humains de la fiabilité dans les systèmes complexes. 1990, p. 223-240.

    There is also a fascinating essay by A. Sabl, “False Frankensteins: the costs and illusions of computer mastery” Techné 2001 v.5 nr.3 p.62-81

  7. Pingback: Some things read this week, 9 - 15 September 2007