TDG Scholar TDGScholar Committed to research! Version 1.7

Help

Introduction

TDG Scholar is a search service that allows to browse a comprehensive bibliography database using a powerful query language. It also allows to export bibliography reports to HTML, BibTeX, and share them with others. The best way to command TDG Scholar is to wander about. Click on the "Bibliography" link in the navigation bar, and try your first queries using a few key words. For instance:

web databases
John McCarthy
Inf. Proc. Lett.
SIGMOD

When you enter a few key words and press enter, TDG Scholar seeks for publications that include these terms, and will show them in (hopefully) a couple of seconds or less. Notice that the terms need not be exact, and that they may appear in any order in the title, the abstract, the journal or conference name, as well as other fields in our database. If you wish to tight the search, just quote your search terms. For instance:

"software engineering"
"information extraction"
"semantic web"
"web wrappers"

Notice that search results can be browsed, i.e., if you click on the name of an author, TDG Scholar will search for his or her publications and will show them to you. Similarly, if you click on the name of a journal, a conference, or a publisher, TDG Scholar will also search for publications hosted by that journal, conference or publisher. Note that you can refine your query as you click if you tick the "Refine on Click" checkbox. For instance, if you search for publications on "software engineering" and then tick the "Refine on Click" checkbox and click on author "Friedrich L. Bauer", you will only get his publications on this topic.

If you whish, you may export the results of your query to HTML or BibTeX, and link the resulting report from your home page or share it with the help of a bookmark manager.

Search services

In the introduction, you have become familiar with searching for bibliography. Searching for writers, journals or conferences is also very simple. For instance, click on the "Writers" link in the navigation bar, and enter these key words:

Edsger W. Dijsktra

What you get is a list of writers that are related to Edsger W. Dijsktra, either because they have co-authored an article with him, or because they have written a paper about him. You may try several writers at a time, and you will get the list of writers who are related to them all. For instance:

Edsger W. Dijsktra and Alan J. Perlis

Give also a try to our search for journals or conferences. For instance:

transactions  
int. journal
image processing
knowledge engineering
IEEE Multimedia
JISBD

Please, notice that the previous queries do not search for writers or journals whose names match the key words you have provided; instead, they search for publications that match those key words, but do not report on them, but on the writers of the journals that are related to them. This feature makes TDG Scholar very valuable. For instance, if you wish to know who has written a paper with Donald E. Knuth on "font caching", try the following query in the writer search service:

Donald E. Knuth and font caching

If you try the same query at the journal search service, you will get a list of journals in which Donald E. Knuth wrote on "font caching".

Query language

Our query language is very powerful, but it is quite simple to learn. You have already tried a lot of examples in the previous sections. For instance, try the following query:

Per Brinch Hansen

This will retrieve the publications that contain words that are somewhat similar to "Per", "Brinch", and "Hansen" somewhere, e.g., in the title, the writer list, the abstract, or even in the name of a journal, a conference, or a publisher. If you wish to retrieve only publications in which "Per Brinch Hansen" is the author, try the following query instead:

author:"Per Brinch Hansen"

author is a search key, and we can handle a lot, e.g., editor, title, journal, conference, publisher, series, organisation, abstract, year, or type, just to mention a few.

For instance, let us search for classical publications by Niklaus Wirth on Pascal. It is as simple as issuing the following query:

author:"Niklaus Wirth" and title:Pascal and year < 1990

What about if you are interested in publications on Modula, as well? There is no problem at all since our query language supports logical operators and, and not, and or. For instance, the following query refines the previous one and retrieves publications on Modula, as well:

author:"Niklaus Wirth" and title:(Pascal or Modula) and year < 1990

Note that search terms are not interpreted literally, but as prefixes. That is, if you issue a query like the following, then you will get publications that include prefix "fair" in their title, which includes terms such as fair, itself, fairness, fairly, or Fairfax, VA, USA.

title:fair

Searching for prefixes is a good technique, but, unfortunately, it falls short in cases in which the term for which you are searching has irregular inflected forms. For instance, if you are interested in papers about "repairs", you might be interested in papers whose title contains terms "repair" or "reparation", as well. In such cases, you need to instruct our search engine to inflect the search term in which you are interested using key word stem, namely:

stem repairs

But, what kind of repairs are you interested in? Maybe you are interested in repairing syntax errors in LR parsers. In such a case, you are looking for the following publications:

title:(stem repairs and "syntax error")

Notice that "syntax error" is written within quotes. This tells our search engine that prefixes "syntax" and "error" must appear in sequence. This is a simple form of proximity, but our search engine allows for fuzzier proximity constraints. For instance, if you wish to refine the previous query so that only publications in which "syntax error" is close enough to "lr", which is a well-known parsing technique, try the following query:

title:(stem repairs and "syntax error" near lr)

Sometimes, you are interested in publications that have been published in, say, journals or conferences. In such cases, you must include a type restriction in your query. For instance, the first query below returns the books by John E. Hopcroft, and the second one his journal articles:

author:"John E. Hopcroft" and type:book
author:"John E. Hopcroft" and type:article

Embedding reports

You might be interested in embedding a report with your favourite publications in your home page. This task is very simple, since you just need to copy the following tiny script at the place where you wish the report to be displayed:

<script type="text/javascript" 
        src="http://scholar.tdg-seville.info/UserArea/Reporter.ashx?query=your-query">
</script>
<div style="font-size: small">
    This report was served by <a href="http://scholar.tdg-seville.info">TDG Scholar</a>
</div>

The only thing you need to do is to replace your-query by the query in which you are interested. Please, do not remove the acknowledgement line; this is the price you have to pay to use this service.

By default, the reports that this script produces are plain text. Should you be interested in endowing it with a little colour, please, consider linking the following CSS Style sheet from the head section of your web page:

<link rel="stylesheet" type="text/css"
      href="http://scholar.tdg-seville.info/UserArea/Simple.css"></link>

Troubleshooting

Please, bear in mind that our services are just a proof of concept. They run on a modest computer that cost us €500, and we are not planning on investing more in hardware in the near future. Due to this extreme hardware constraint, we restrict searches to 100 publications, writers, journals, conferences, or publishers. This should be enough for average users.

We implement a highly optimising index-based searching procedure. It deals very well with popular queries, but takes a little more than expected with unusual queries. In such cases you are likely to get a time-out message. The solution to this problem is very simple: just wait a couple of minutes and try it again. The first time you submit a query, our search engine analyses it and devises an execution plan that is reused every time it gets a similar query. Even if you get a time-out, our search engine keeps working to serve your query. Thus, if you try it a little later, our search engine will retrieve the execution plan from our database and will serve your query as efficiently as possible.

Searching for authors, journals, or conferences is a little more difficult than it might be expected. Please, recall that when you issue a query like "Butler W. Lampson, we are not seeking for authors with a similar name, but for authors who are related to him, as well. The algorithm that implements these queries is complex, but recall we need to analyse your query only once; the second time it is executed, the execution plan is fetched from our database and it is executed very efficiently.

Sometimes, you see a message like "Search took 0.156 seconds", but you feel that the search has actually taken more time to complete. Note that we heavily use XML and XSLT transformations to render the user interface; unfortunately, these technologies usually take a little more time than would be desirable. The previous message means that our system took 156 milliseconds to execute your query, but it does not account for the time required to render the interface or the network latency.

If you happen to find other problems, we would appreciate if you could, please, take the time to report them to us. Please, send a message to contact at tdg-seville.info and describe the problem you've found as precisely as possible. Thanks!