PROF. M.R. KUMBHAR	PROF. D.B. KRISHNA RAO

But now there are such attempts called "Semantic Web" attempts. .  It is not possible because web is growing every second and chaotically too. Freedom of expression and the democratic movement has taken root in every educated individual irrespective of the ideology followed in individual societies. If it is forced, then the Japanese Garden web type documents will equal the number of such gardens existing in the real world, for the rest  would  turn to blogs, myspace, e-mail to web document and such easy web publishing,  leaving the Japanese garden type web documents to the "elite" and the rich. The web will be as chaotic as ever and special search engines such as blog search engines, email search engines and rss feed search engines etc., would become the order of the day.

Semantic Webbers would like to keep the forest, but make it a processable one !. They would like the unstructured or ill structured documents to be well structured. Is that achievable? OK you use RDF, XML, Ontologies (It is a faceted classification scheme). . At present it has been proposed that the ontologies must be specific domain based. This is actually again like  Librarians' method of managing the information.  Librarians knew that it would be difficult to manage one library collecting and organizing everything and made subject specific Libraries and Information Systems. You have the National Library of Medicine, the National Agricultural Library and so on. Now the semantic Webbers want to follow this model and suggest that ontologies must be domain specific !!

Alright, you have domain  specific ontologies/ faceted classification schemes.  All the web documents on the web are put as RDF using OWL, ontologies and all that stuff. That is, you have all the metadata and all the agents.  . So what? They are processable, you can find out which medical doctor is available at a time suitable to you and all that stuff, PROVIDED YOU IDENTIFY FIRST THE WEB DOCUMENT HAVING SUCH TYPE OF DATA. . But how are you going to identify which document or web documents have to be accessed to further process by your agent. Is that also going to be identified by your agent? Are you going to have agent of agents to select the appropriate agent? Is your agent going to traverse the web to find out the documents to be processed? Is it that for a query the super agent to which the query is submitted selects the specific special agent / agents. The specific query using ontologies is processed and then the agent enters the web. But where? Into the forest and starts processing all the processable? You are going to have trouble as you may have to crawl at least a good part of the web, and follow the links provided in the web documents identified, to get the right data to process. .What to do then? You have to have an index to the web which you can search first then select the best fit from it, identify whether the required type of data is available in it and then process it.

When you wish to traverse a desert you perhaps cannot have a map  and rely on it. But you may be able to use a compass. However  the early explorers did a great job of developing maps for the routes they undertook. Not only they developed maps of the routes, they also kept diaries of the customs of the places they visited, the dangers, the wise things to do and so on. They had maritime maps and complicated codes using stars and planets and so on.  Just like the early explorers set out to explore the earth and prepared maps to indicate what is where, what precautions one must take to be there etc., a map of the web must be constructed. The picture on the right hand side is not the appropriate one. I saw a cartoon perhaps drawn by R.K. Laxman after he visited  Los Angeles during the  1980's and got lost in the mesh of roads, while he ventured to explore the terrain by walking.  That would have been the most appropriate. It was perhaps published  in Span or The Hindu I am not sure. If any one comes across a weary backpack man on foot in front of a map of a flyover on the side of a flyover looking at the sign "You are here" please let me know.

The possible surrogate systems are systems like Google, Yahoo, etc. These have already built up the necessary tools. But what is lacking is the inclusion of a facet analyzed subject heading with appropriate superordinates to each of the components in the heading enriched with synonyms.  The required classification and indexing tool called Classaurus can also be derived from this facet analyzed subject heading.  The facet analyzed subject heading / POPSI heading (for style sake it can be referred to as Logico Semantic Domain Expression (LSDE) -- which is nothing but a structured subject heading), may have to be assigned to each of the meaningful units / sections and subsections of the web document as required.

If by developing complicated standards, putting a page on the web becomes like touching the nose by winding the hand around the back of the neck, then the freedom loving, line of least action takers, would just put their web documents as blogs, e-mail (HTML), myspace, discussion lists (HTML) and such other easy, quick methods not caring for all the high flown standards of RDF, OWL, Ontology (Faceted Classification scheme) etc.. Then the web would become more chaotic than  it exists today. However, any processable data / database in the individual web document could be indicated in the surrogate with structure identification , if necessary a sample of the data, perhaps one row of the data from the table in the web document with datanames (meta data) could be put in the surrogate for easy identification of processing routines. Better still, any routine necessary for processing could be stored in the surrogate itself. But for solutions that require data from different documents to be merged, some generalized processing routines would be required .

LET ME RECAPITULATE

1)Because the facet analyzed POPSI (POstulate based Permuted Subject Index) headings can produce an organizing effect while sorted, each of the web documents and their worthwhile sections and subsections, coherent text, or image or audio or video or any combination of these, must be fitted with such a heading, having, a Base or Discipline, its divisions and subdivisions as applicable, the Main Object or Core Entity, its species, parts, constituents, any Property of or Actions on the object or by the Object, and its types followed by Common Modifiers, each one of them "modulated" by their respective superordinates enriched with synonyms. If these are sorted then that would constitute the World Wide Web Map (WWWM), exhibiting an organizing effect. [Please see Bhattacharyya's POPSI : Its fundamentals and procedure based on a general theory of subject indexing languages, in "Library Science with a slant to Documentation, Vol 16 (1979) Iss (1); pp 1-34" which is not available on the web. However you can access this <http://drtc.isibang.ac.in/~guha/popsi/popsi-doc.pdf >, you may not have this "International Federation for Documentation / Classification Research Report No 21:  Computerized Deep Structure Indexing System, Indeks Verlaag, Frankfurt 1986" but you can have a look at "Online Construction of Alphabetic Classaurus" </devadason.geo/OnlineClassaurus.htm >, and of  course  http://us.share.geocities.com/devadason.geo/DSIS.pdf 

The most important classic document  from which all the Facet Analysis stalwarts -- those who have developed "facet analysis for dummies", "easy facet analysis", the "true and simple facet analysis" and brought fame to them  -- have copied is "Prolegomena to Library Classification by S.R. Ranganathan" and is available at <http://dlist.sir.arizona.edu/1151/>  This is edition 3 1967. Some how I have a fascination for edition 2, 1957.

2) Such headings (Logico Semantic Domain Expressions - LSDE's) could be formed easily by the web document builders as it is almost similar to forming an expressive title to the web document, which can be done by answering a set of questions and following a few guidelines. Or it can be done by the surrogate creators / search systems.  Initial efforts would become less time consuming as the web maps get built and are made available for reference while creating the subject headings.

3) These subject headings could be used to form the Classaurus (faceted classification scheme with vocabulary control features) and be used for easy translation of faceted  headings from one language to the other (there are some problems, due to certain concepts not existing in certain languages and terms to denote them being not available and so on), and to categorize the web documents.

4) It would be worthwhile exploring the possibility of utilizing the existing library system such as the different National Libraries,   having the necessary expertise in particular fields, to create and maintain specific WWW Maps for the areas of expertise of individual national libraries. For instance, the  National Agricultural Library could be used as the agency responsible for mapping any web document to be categorized as belonging to Agriculture. The agency has a well developed thesaurus and would be the most appropriate agency for developing the Agriculture WWWM. This agency could be assigned a range of IP addresses (covering a specific geographical area or so),  to monitor and update the web. In a similar way the existing information handling expertise could be channelized to form National WWW Maps in different subject areas and merged. It could be formed language specific, but there should be an English language LSDE for every subject heading in the summary record / surrogate. Switching between languages may be possible, but there are problems when such modulated subject headings are translated from one language to another. Even hierarchy would be a bit difficult to map correctly !. I do not want to go into examples here.  The allocation of work could be on the style of cooperative global information system, and it could be subject - specific, language - specific, nationality - specific with full coverage of  designated  IP address ranges of the web documents assigned to individual agencies to build these maps and maintain them avoiding any duplication of effort.

5) Any speific data or database included in the web document could be indicated in the surrogate for that web document, with a model of it or structure of it, or even an example of it consisting of one row of the data with attached datanames / meta data. This will help prepare the processing routines for further processing of the data available in the web document.  It would be even better to store the processing agent in the surrogate itself or provide a link to it, so that it can be loaded for processing.

21 May, 2007.