A web-based educational library

Abstract - We present an Internet library, built by the authors, specialising in managing contents, which have been previously selected from the net. The user-teacher will have a tool to classify his preferred learning materials found on the web. The user-student will have a friendly interface to find the needed contents guided by the teacher.

I. INTRODUCTION

Most search engines currently present on the world wide web (WWW) [1] are usually pointing to commercial activities. When the user introduces a word in the corresponding text area of the searcher's home page, it makes an indiscriminate search among thousands of pages. Often, the criteria used to offer the searching results is a commercial one. Recently some efforts have been made to overcome this indiscriminate search by means of human intervention. At this moment all searcher's home pages offer, through directory-based interface, a selected set of webpages[2,3]. This selection is achieved by some algorithm that uses parameters, like the frequency of users visiting the pages, mixed with human intervention.
These generic search engines have influenced others in the educational field. However, an educational search engine must be something close to a library.
Keeping this problem in mind, and aiming to solve it, we have recently developed an internal algorithm, beside a user interface, which allows teachers and students to use the WWW in a non-indiscriminate way. We have named the program which implements the algorithm and the interface, netbrary. This word comes from the words net and library, following the sequence: library, li-brary, net-brary, netbrary. Using the authors language, which is Catalan, it is named, xarxateca: biblioteca, biblio-teca, xarxa-teca, xarxateca. The netbrary is at present a part of the educational project la baldufa[4,5,6 ].
In the following sections we will show how the netbrary works, internally and externally. The text included in examples and figures is in the original in Catalan, but we have translated it in the corresponding figure text when necessary.

II. BASIC PRINCIPLES IN THE NETBRARY'S DESIGN

A.. There is human intervention in selecting the materials introduced in the netbrary

This is an obvious requirement. In some sense, it is part of our project's objective to reject all kind of automatic selection. The netbrary covers the needs of a single educational institution. We are thinking about 100 to 10000 items to be searched, not milliards!! If a webpage is not interesting for us or it can confuse students we simply do not include it on the netbrary. Of course, in the process of selecting pages we can use the conventional searchers, and, after that, catalogue the webpage founded. For example, with 16-years, physics, mechanic, dynamics, wording problem a page which talks about "....a Ferrari is driving round a curve on Louis Armstrong street...What is its normal acceleration?..." the words Ferrari and Louis Armstrong are not describing anything essential on this page.
On the other hand it is important to note that the new cataloguing of a webpage is always possible. The person responsible for the netbrary can introduce new items each time it is necessary.

B. The process of cataloguing the material is made using keywords and conceptual sets of keywords (CSK)

The netbrary needs each item to have some pre-selected keywords. The keywords have two faces. One that is for internal use, and other one is which the user will see. Single keywords can be joined building a conceptual set of keywords (CSK). [law, newton], which refers to Newton's Law, is an example CSK. This is important because, if the user searches, for instance, some Smith's Law (suppose that Smith has many laws) related to ecology, we will ask to the netbrary for :[law,smith],ecology. Note that the netbrary knows that he/she doesn't want all laws of ecology. The CSK [law,smith] is a single not separable concept. Later we will see other advantages of using CSK.

C. The user-student may not know exactly which keywords she/he must introduce when searching

D. The result of a search may give some interesting information about the document found

The result of a search is a list of URLs related to the CSK required by the user. Some information about each URL may be offered in order to make it easy for the user to choose a suitable document. Conventional searchers give either the first paragraph of the page or the paragraph where the keywords appear. Unfortunately this doesn't give information about the web page's real content. Instead of that, the netbrary gives the information previously introduced in the cataloguing process. It uses also the CSK structure to give, when possible, information relating the user's inquiry. For example if the user is asking for some law in ecology: [law],ecology, and if the netbrary finds some documents and one of them is catalogued as [law,smith], this information will appear in the list of results. That is, the netbrary may be able to relate the user's inquiry with the extra information offered.

E. All the processes of searching can be done without net-connection.

This will allow the users to perform searching sessions or take advantage of navigation without net connection. The connection is only necessary when she/he wants to reach some of the documents listed in the searching results.

III. THE NETBRARY INTERFACE

The interface of the netbrary has been size-optimised. It initially appears at the bottom of the browser's window and can be visible simultaneously with the content document. The user can consult the documents found without losing either the list of results or the configuration of selected words used in the process.
In Fig. 1 we show the area occupied by the netbrary . This version is the most complex and powerful that we have designed.

Other versions can take up less vertical space. Nevertheless, when the user is not using the netbrary, she/he can drag the frame bar with the mouse to diminish the frame size occupied by it. When she/he needs it she/he can resize the frame by dragging again the frame bar with the mouse.
At the top of the netbrary there is a text area where the user puts the keywords (see Fig. 2). But the user doesn't need to know which keywords are better to write. A set of dynamic lists offers to the user all the keyword (not forced, see the paragraph below) available. The keywords are distributed thematically through the lists. The lists can be either visible or hidden by clicking on the corresponding buttons placed on the left of netbrary window.
There are two kind of keywords: forced and not forced. The first ones are typically keywords referring to the contents level, the kind of document required (theory, problem, activity) and general area of knowledge the user wants to work on. The forced keywords are accessible by the rolling lists that appear on the left of the window. In Fig. 1 and 2 we have chosen :
*la baldufa: which means documents in la baldufa server.
*Teoria: which means theory like documents.
*Nivell de dificultat: Difficulty level. We have not forced any level
*Electromagnetisme: Electromagnetism
Automatically the netbrary offers through the lists all non-forced words related to existing documents having the selected forced words.
The lists have some hierarchy, from the left to the right. If we select some topic in the first list, as we show in Fig 1., where we have selected Electrostatica (which means electrostatics), all other lists filter their contents offering only keywords related to documents that have also electrostatics as a keyword.

In Fig 2. we have selected Camp (which means field) besides a double clicked to put this word into the text area.

Automatically the netbrary opens some extra lists showing the words that joined to Camp forms a CSK. That is, the netbrary informs us that there are actually some documents catalogued with these CSK, and we are free to form the desired CSK by double clicking on the words to put them into the text area. It has no sense to try to form CSK not offered by the netbrary simple because the non offer means that at the moment there are no documents related to these CSK. We can finish the CSK by clicking the F button on the top-left corner of the netbrary window. After we have formed the set of desired CSK we can click on the C button on the top-left corner and after that the searching process begins. The result will be showed in a layer at the same place. We can consult the results by clicking on them. The document will be showed in the window just above of the netbrary.
Suppose we want documents about Faraday. We only have to select Faraday from the corresponding list and begin the searching process. In the result, which shown in Fig. 3, we can see that the CSK related to Faraday are showed as an interesting information related with to each item found.

IV. INTERNAL ASPECTS OF THE NETBRARY

The netbrary is a Java applet so it can work exclusively on the client-side. As all applets it is incrusted on a web page by using the common tag <applet></applet>. There are some PARAM available to configure the size and other common aspects of the interface. We have included on the page some Java Script (which communicate the applet to the page) in order to optimise the interface. Internally the netbrary is configured basically using two files: the keyword-file and the url-file . In Fig 4. we can see a piece of those files. The keywords-file is where all keywords are declared. Each keyword has an internal form and a public form. This is very useful mainly for two reasons:
* It allows to use some special characters (spaces, accents...) in the public version
* It allows to easily translating to other langauge without much work.

Each keyword is attached to a category, which is visible in the interface through the different lists (for not forced categories) or through rolling lists (for forced categories).
The url-file is where the url addresses are declared. Each time a new document is entered in the server, or a new url must be included in the library, the netbrary administrator must edit this file to declare it. It is formed by sets of three rows. The first row is formed by the url, which can be an html page or an image or a *.doc file... etc. The second row is the visible comment that will appear in the result. The third row is where the CSK are related to the url. The CSK are formed by keywords previously declared in the keyword-file.

IV. ACKNOWLEDGEMENTS

X.J. would like to thank the Comisión Asesora de Investigación Científica y Técnica for partial financial support, under Contract No. BFM2000-0604 and the Laboratori de Física Matemàtica at the Societat Catalana de Física for partial financial support.

The authors would like to thank the Departament d'Universitats, Recerca i Societat de la Informació (DURSI), de la Generalitat de Catalunya for partial finantial support.

V. REFERENCES

1. http://www.google.com/
http://www.yahoo.com/
2. http://www.google.com/dirhp?hl=us
3. http://directory.google.com/Top/Science/Physics/Relativity/
4. http://baldufa.upc.es/recerca/indexen.htm
5. X. Bohigas, X. Jaén and M. Novell. "Teaching and Learning Physics using the Internet: la baldufa Project". Higher Education in Europe, Vol. XXIII, No. 2, 1998 (233). http://baldufa.upc.es
6. X. Bohigas, X. Jaén, M. Novell "A web-based project to learn and teach physics (la baldufa)" Proceedings International Conference Physics Teacher Education beyond 2000. August-2000. Barcelona..