Deprecated: Assigning the return value of new by reference is deprecated in /var/www/thorendal.dk/public_html/bbclone/lib/io.php on line 151
Strict Standards: Only variables should be assigned by reference in /var/www/thorendal.dk/public_html/bbclone/lib/io.php on line 154
Deprecated: Function eregi() is deprecated in /var/www/thorendal.dk/public_html/bbclone/lib/new_connect.php on line 88
June 6, 2005
What is relevance?
Deprecated: preg_replace() [function.preg-replace]: The /e modifier is deprecated, use preg_replace_callback instead in /var/www/thorendal.dk/public_html/wp-includes/functions-formatting.php on line 75
To begin with I have to say, that I will not be claiming to talk the truth about relevance. I do though find it very interesting - and puzzeling - that relevance can be view upon from so many different perspectives. And that is why I will write some posts about different aspects of relevance.
I have chosen to call this post:
What is relevance?
The concept of relevance differs depending on which article or book you read on the topic. Today it is more or less agree upon that relevance is a dynamic concept, in the sense that it is a mixture of classic algorithmes for relevance and the users own perception of the concept that constitutes the concept.
Therefor I will have a look at Saracevic’s (1996) types of relevance, as they do take into account that relevance is a dynamic concept. I like his way of differentiating types of relevance and mixing the objective and subjective types. That does not mean that others haven’t comitted good theory on the topic. I just happen to know and like Saracevic.
Saracevic divides relevance into five types:
1. Algorithmic relevance. This is an objective type of relevance where the outcome totally is depending on the match between query and the collection. The result is based on algorithmes and the degree of succes is reliant on the degree of consistency of query and reitrieved documents.
2. Topical relevance. This kind of relevance is very much depending on the indexing policy that the system is based on, as the relevance is compliant with ‘aboutness’. Relevance is only seen as topical matter and is therefor not necessarily reliant on the document. The succes of this kind of relevance is therefor very much dependent on the users ability to formulate their actually needs (the request) and on the retrieved documents.
3. Pertinence or cognitive relevance. This is a subjective kind of relevance. The relevance is to be judged in accordance to the actual information need. The succes is depending upon the indexing policy and methods and the users ability to understand it’s own needs and ability to formulate these needs up against the system. Pertinence or cognitive relevance is the relation between the user’s perception of the need and the retrived documents.
4. Situational relevance. This is a subjective and dynamic type of relevance, which is relying on the user’s perception and interpretation of the tasks and the social-cognitive context. That means that the concept situational relevance is the relation between the retrieved documents and the user task as the user interpret them. So the succes in situational relevance is very much depending on the users ability to use a piece of information to satisfy it’s need.
5. Motivational or affective relevance. This is a completely subjective type of relevance. It is a bit vague what this kind of relevance is besides it is the degree of satisfaction, succes or the like that can affect the user’s motivation and opinion.
It has to be remebered that every time the talk is on relevance, the context is very often ignored for the fairly static definitions of different types of relevance. The user’s inherent information is taken from one context and has to be used in another - the system. Relevance is multi-dimentional and complex, but if the user’s perception of relevance is taken into account, then relevance as concept can be useful.
In general the cognitive approach to information retrieval is gaining ground as systems are becoming more and more complex and the standards are raising. Earliere on the complexity didn’t excist to the same extent as today. More systems are coming into the everyday life and more users will be using these systems. Therefor it is important that the types of relevance that are used reflect this. So even though Saracevic’s types of relevance are fairly simple they are easy to use as meassurements for different kinds of more or less binary relevance.
References:
- Borland, P. and Ingwersen, P. (1998). Measures of relative relevance and ranked half life: Performance indicators for interactive IR. In: Croft, B.W., Moffat, A., van Rijsbergen, C.J., Wilkenson, R., and Zobel, J. eds. Proceedings of the 21st ACM Sigir Conference on Research and Development of Information Retrieval. Melbourne, 1998. Australia: ACM Press/York Press, p.324-331.
- Borland, P. (2000). Evaluation of interactive information retrieval systems. Åbo: Åbo Akademis förlag. 276 p.
- Saracevic, T. (1996). Relevance reconsidered ‘96. In: Ingwersen, P. and Pors, N.O. (eds). Informations Science: Integration in Perspectives. Proceeding of CoLIS 2, Second International Conference on Conceptions of Library and Information Science, Copenhagen, 1996. Copenhagen: Royal School of Librarianship, 1996, p.201-218.
- Schamber, L., Eisenberg, M.B. and Nilan, M.S. (1990). A re-examination of relevance: Toward a dynamic, situational definition. Information Processing & Management, (26), p. 755-775.