Documents

Despite the undoubted resourcefulness that users display for making a little information go along way, inferences based upon URLs are always likely to be an unreliable way of making judgements of document content. Instead, it is meta-data that is the key element in improving users' capacity to make informed assessments of a document's quality and relevance [2]. Of course, the form this meta-data will take must be dependent upon the kind of document it describes. Whilst it is easy to specify that the meta-data for a scientific paper would include its authors, key words and abstract, that of many other document types is more difficult to anticipate. The issue that must be addressed then is how such meta-data may be effectively presented to the user. One possible approach is to use meta-data to provide some insights in to the document structure. A simple illustration of how this might be done is shown in Figure 5. Here we see the document content summarised by:

  

Figure 5: An example of document meta-data: structure and content summaries.

1.

Author, title, keywords, abstract and section names appearing in the document. HTML already provides some support for document meta-data through the use of a number of pre-defined tags. These include author, title, and keywords. The new proposed document markup language XML goes further in that it allows the author to define new tags; in fact, every component of an XML document may be tagged in this way.

2.

The distribution of keywords through the document body. This can provide a better indication of relevance than a simple count [11].

3.

A thumbnail outline of the document. The concept of genre is one possible way in which broader notions of meta-data may be addressed. The use of genre in the print media, for example, enables people to make quick and easy distinctions between the vast variety of publications available. To help the readership distinguish between news genre, simple visual cues such as the tabloid and broad sheet format have been developed. The thumbnail is an attempt to provide similar visual cues for Web documents.

4.

The numbers and sizes of different embedded media types and the number of links. This also may provide the user with genre cues. For example, a document containing many links may appear to be a richer resource for browsing.

  

Figure 6: An example of link QoS data: channel and site QoS summaries.