IT Services



Oxford Google Search Appliance Documentation


Contents

Oxford University has obtained a Google Search Appliance to facilitate improved site searching across the Oxford ox.ac.uk domain. This now replaces the previous Google University search which has been in use for the last few years. The internal search remains HtDig for the time being, but may be replaced by the new search appliance in due course.

The new appliance is being maintained and managed for the University by OUCS's InfoDev Team. If you have any problems with the search appliance you should contact this team either using the OUCS Feedback form or via the email address infodev@oucs.ox.ac.uk. The following sections describe how to use the new search engine on your own Oxford site.

The appliance is at http://googlesearch.oucs.ox.ac.uk



1. Indexing Policy

The Google Appliance crawls across Oxford continuously to update the search index. This ensures the search results are up-to-date. For a site to be indexed by the google appliance, pages must:

If you web site does not meet these criteria, but you would like to see if it can be included, please contact InfoDev Team for advice. If your web site does meet these criteria but does not appear in search results, please contact InfoDev Team.



2. Excluded web servers

There are a small proportion of web servers that are not currently indexed by the search engine because they cause looping, generated infinite results etc.

We have also excluded all URLs that have a question mark in the URL e.g. http://www.webserver.ox.ac.uk/index.asp? or index.php? etc. If you have a site like this and need to have it crawled, please contact InfoDev Team. Other types of pages excluded are any that cause repeated visits such as online calendars etc.



3. Adding the Standard Oxford Google Search to your site

Pointing your site search to the new Google Appliance is a straightforward procedure. To apply an Oxford-wide search to your pages you need to add the following code (this is in XHTML):

<div class="searchbox"> 
  <form id="searchform" method="get"
	action="http://googlesearch.oucs.ox.ac.uk/search"> 
    <fieldset>
      <legend>Your text or image here</legend> 
      <input type="hidden" name="as_sitesearch" value="ox.ac.uk"/>
      <input name="UnitPicture"   type="hidden" 
	     value="http://www.oucs.ox.ac.uk/googlesearch/radcliffe.jpg"/>
      <input name="client" value="oxford" type="hidden"/> 
      <input name="proxystylesheet" value="oxford" type="hidden"/> 
      <input name="output" value="xml_no_dtd" type="hidden"/> 
      <input name="q" id="input-search" size="25" onfocus="clearsearch();" 
	     maxlength="2048" value=""  type="text"/>
      <input name="Go" value="Go!" type="submit"/>
    </fieldset>
  </form>
</div>

Figure oxford-search.gif [Oxford-wide search results]



4. Adding a customised Google Search to your site

The Google appliance has the ability to allow customisation of the search form and the results it sends back. This allows searches to be restricted to a given domain e.g. oucs.ox.ac.uk. In order to take advantage of this facility a few of the parameters set in the Oxford-wide search need to be altered. The code lines that need to be amended are:

  1. <input type="hidden" name="as_sitesearch" value="ox.ac.uk">
  2. <input name="UnitPicture" value="http://www.oucs.ox.ac.uk/googlesearch/radcliffe.jpg" type="hidden">

The first input box needs to be replaced with:

<input type="hidden" name="as_sitesearch" value="your url">

This will make the search look at your own URL instead of being an Oxford-wide search.

To restrict the search to Physics for instance, the input code should look like:

<input type="hidden" name="as_sitesearch" value="physics.ox.ac.uk">

You may restrict the search even further by adding a trailing / to the end of the url. The search is then restricted to only that folder.

e.g. <input type="hidden" name="as_sitesearch" value="physics.ox.ac.uk/admin/">

You should then insert the following code:
<input name="Unit" value="Your Department Name Here" type="hidden">
This input box needs to have the correct department name added. So, for the Physics Department, you should replace 'Your Department Name Here' with 'Physics' as the value parameter. The name Physics will now appear as part of the results page heading.

Finally you can change the image that appears to the left of the search box. Change the URL so that it references your chosen image instead of the default Oxford image. Please make your image no wider than 75px to prevent CSS problems. Test your changes to make sure that you now see your chosen image, the search title states your unit name and finally that the results are coming from your chosen Oxford domain.

Figure physics.png [Physics search results]

Please note: You cannot specify more than one domain using the as_sitesearch parameter. However multiple web sites within that domain are all searched. For instance, public facing pages of both www.magd.ox.ac.uk and jcr.magd.ox.ac.uk can be searched together using this method but www.magd.ox.ac.uk and oucs.ox.ac.uk cannot. If you wish to include more domains in your search results a search 'collection' needs to be used. 'Collections' are created for you by the InfoDev Team. See 5. Google Customised Search Collections for more information.



5. Google Customised Search Collections

Search collections are made up of one or a series of URLs and given a group name e.g. Chemistry. They allow you to specify what constitutes your search and means you can taylor the search results to one or a small group of URLs. If you prefer you can use the generic Oxford collection instead of creating one for your own unit. If you require a collection to be set up for you, please contact the InfoDev Team. The following code sets up a Oxford-wide collection:

<input name="site" value="oxford" type="hidden">

If you are using your own collection, the value attribute needs to change to the name of the collection you have asked InfoDev Team to create for you. The OUCS collection is called oucs. If you have not requested a collection to be set up for you , but just want to be able to search the whole of Oxford University, keep the value as oxford.

Please note: you can only use one collection name in your search.

 <div class="searchbox"> <form id="searchform" method="put"
               action="http://googlesearch.oucs.ox.ac.uk/search"> <fieldset>
               <legend>Your text or image here</legend> <input
               name="site" value="oucs" type="hidden"> <input name="Unit" value="OUCS"
               type="hidden"> <input name="UnitPicture" type="hidden"
               value="http://www.oucs.ox.ac.uk/googlesearch/radcliffe.jpg"> <input
               name="client" value="oxford" type="hidden"> <input
               name="proxystylesheet" value="oxford" type="hidden"> <input
               name="output" value="xml_no_dtd" type="hidden"> <input name="q"
               id="input-search" size="25" onfocus="clearsearch();" maxlength="2048" value=""
               class="vam" type="text"> <input name="Go" value="Go!"
               type="submit"> </fieldset> </form>
               </div> 

Figure oucs-search.gif [OUCS collection search results]

To request the creation of a collection, please contact the InfoDev Team with details of which domains should be included in your collection.



6. How to obtain XML formatted search results

On occassion you may require XML formatted search results when you do searches. You may also obtain the Document Type Definition (DTD) if you set the output parameter xml instead of xml_no_dtd. In order to do this type of search you should set up your web form in the following manner:

 <div class="searchbox"> <form id="searchform" method="put"
               action="http://googlesearch.oucs.ox.ac.uk/search"> <fieldset>
               <legend>Your text or image here</legend> <input
               name="site" value="oucs" type="hidden"/> <input name="client"
               value="oxford" type="hidden"/> <input name="output" value="xml"
               type="hidden"/> <input name="q" id="input-search" size="25"
               onfocus="clearsearch();" maxlength="2048" value="" class="vam" type="text"/>
               <input name="Go" value="Go!" type="submit"/> </fieldset>
               </form> </div> 

Here the Unit, UnitPicture and proxystylesheet inputs have been removed, and outputnow has a value of XML.

Performing a search with the above form will present the search results in XML format:

Figure xmloutput.png [XML search output]

The returned search data can now be formatted using your own customised XSL stylesheet and finally presented to the end user.



7. Synonyms

Synonyms can be created to suggest alternate search queries to perform. An example of such a query would be if the name of someone, or indeed something, has one or more alternatives e.g. "President Jones" could be a synonym for "Fred Jones" suggesting that users may make alternate queries for one topic. The use of such synonyms will help users find the information they are seeking. Please contact OUCS's InfoDev Team if you need any synonyms creating for your users.