Federated Query Server

Frequently Asked Questions

Some frequently asked questions about the OpenText Federated Query Server product

What is Federated Query Server?

Federated Query Server is a piece of search middleware that takes a user's query, translates and broadcasts it to multiple HTTP-enabled content repositories, reformats, sorts, merges and clusters the query results, and returns them to the user in a single page.

Which repositories can be searched by Federated Query Server?

Federated Query Server is compatible with the following OpenText products: Content Server (formerly Livelink), Media Management (formerly Artesia DAM), Discovery Server (formerly BRS/Search) and Collections Server (formerly BASIS).

It has also been tested against many other Web-enabled Intranet products, such as SAP, Lotus Notes, Microsoft SharePoint Server and Novell GroupWise - please contact us for further details.

Federated Query Server has been tested against many Internet search engines, including Bing, Facebook, Google, Twitter, Wikipedia, Yahoo and YouTube.

How do I know if a repository is compatible with Federated Query Server?

Federated Query Server requires a content repository to have the following features and characteristics:

• The repository must have an HTTP interface that uses either the GET or POST method
• A query is submitted and the results obtained by visiting one or more URLs
• The query language of the repository is either Boolean or keyword
• The results are returned as a list of items, each with a URL that links to the result document

Any content repository that meets the above conditions is likely to be compatible with Federated Query Server.

Is Federated Query Server OpenSearch compliant?

Yes, Federated Query Server can act as an OpenSearch server, an OpenSearch client, or both. It ships with an OpenSearch description document, and RSS and Atom interfaces. It's configuration tool has built in support for reading OpenSearch description documents, and built in parsing rules for RSS and Atom.

How do I issue a search to Federated Query Server?

By executing Federated Query Server (usually by submitting a form in an HTML search page) and specifying the search criteria in form variables.

Searches can also be issued from the search bar of most browsers, and from Windows 7 Explorer.

What form does a query take?

Federated Query Server accepts Boolean queries. The text of each Boolean operator is configurable, and operators may be disabled if required.

Federated Query Server also supports metadata queries, both fielded and numeric. Again, the text of each metadata operator is configurable, and operators may be disabled if required.

How does Federated Query Server translate queries?

The text of each Boolean and metadata operator, and the characteristics of the language, make up a language configuration. Federated Query Server has one language configuration for the user interface (i.e. for queries specified on the search form) and one language configuration for each content repository.

Any operators that are found in the query text are translated from the user interface language to the languages of the content repositories. If an operator cannot be translated because it is not defined in the content repository language configuration, then either an alternative operator will be used if available, or the operator will be degraded until it can be translated.

How does query operator degradation work?

The following operators are degraded to more general operators if they cannot be translated:

Boolean XOR	⇒	Boolean OR
Boolean same paragraph	⇒	Boolean AND
Boolean same sentence	⇒	Boolean same paragraph
Boolean proximity	⇒	Boolean same sentence
Boolean phrase	⇒	Boolean proximity

Query operator degradation is used in conjunction with looking for alternative operators. For example, if an "alternative Boolean phrase" operator is found in the search form query text, but only a "Boolean AND" operator is defined for the content repository, the "alternative Boolean phrase" operator will be degraded as follows:

Alternative Boolean phrase	⇒	Alternative Boolean proximity
	⇒	Alternative Boolean same sentence
	⇒	Alternative Boolean same paragraph
	⇒	Alternative Boolean AND
	⇒	Boolean AND

Does Federated Query Server support forms with multiple query boxes?

Yes, Federated Query Server can handle search forms with up to 50 query boxes. Each box may have a default Boolean operator that will be used between words where no operator is explicitly specified, and each box may be joined to the next by a conjunctive Boolean operator. Both types of operator are defined in form variables.

What about content repositories that can only handle one query box?

Federated Query Server can either pass the contents of each query box to a content repository, or it can combine the contents of all the query boxes into a single query string and pass this instead.

Additionally, Federated Query Server can extract all the query terms from all of the boxes and pass this also (this is useful for specifying terms for ranking).

Does Federated Query Server broadcast queries in parallel?

Federated Query Server starts a new thread for each content repository. The thread translates the query, connects to the repository's host, issues the HTTP request with the query and any other required information, and collects and sorts the query results. All of the threads are started simultaneously and therefore act in parallel. The main thread is responsible for merging the query results and returning them to the browser.

In what order are query results returned?

Federated Query Server has three types of content repository - the master repository, one or more sorted repositories, and one or more unsorted repositories. The page returned to the client is effectively the page returned by the master repository, with the query results returned by all other repositories inserted into it.

The results from the master repository and all sorted repositories are sorted and merged into a single list which is output first. The results from all the unsorted repositories are output after the merged list.

How does Federated Query Server sort query results?

An unlimited number of sort levels may be specified. Each level may sort on any of the elements of a result - for example, relevance score, title, date, size, category, etc.

Usually, the first level sort is by relevance score, as this is the order in which results are normally returned from content repositories.

Can I cluster query results into groups?

Federated Query Server can cluster query results by their content, their location (extracted from each query result's URL), or both. Clustering by content produces groups of results listed under phrases that appear frequently in the result's title or summary.

Are duplicate query results removed by Federated Query Server?

Duplicate query result removal is optional. If it is enabled, query results that are the same (you can choose which comparison method to use) are merged into a single result, the relevance score of the merged result being the average of the relevance scores of the original duplicate results.

What about different ranking algorithms in different content repositories?

Each content repository returns relevance scores that are calculated using a certain algorithm, and that lie within a certain range of numbers. Federated Query Server can be configured to accept relevance scores in any range (for example, 0-1 or 0-100). All relevance scores are converted to a common scale of 0-100 for internal calculations, but may be output again in any scale.

Federated Query Server can also modify relevance scores returned by a content repository - one of four algorithms may be selected to carry out this modification (for example, one of the algorithms reduces high scores but increases low scores).

What about content repositories that do not return relevance scores?

Federated Query Server can estimate a query result's relevance from its position in the results list. Three algorithms are provided to perform this estimation.

Alternatively, Federated Query Server can retrieve the document pointed to by a result's URL, and assign the result a relevance score based on the content of the document. This feature is very powerful, but will also significantly increase the amount of time before results are returned to the user.

Finally, Federated Query Server can create a pseudo-document from the textual fields in a result (for example, the title and summary) and then assign the result a relevance score based on the content of this pseudo-document.

Does a content repository have to return results as an HTML page?

No, Federated Query Server can accept results in HTML, XML, JSON, CSV, plain text, or virtually any other textual format - as long as the results, and the elements within each result, are identifiable (for example, by strings either side or the presence of an XML tag).

How does Federated Query Server handle authentication?

Federated Query Server supports Basic, NTLM, Kerberos and delegated HTTP authentication. It can also access content repositories that require a user to log-in via a Web form.

Credentials can be stored in the Federated Query Server configuration file, specified on the search form, or obtained from the Web server.

Additionally, Federated Query Server can pass cookies between the browser and a content repository, in either or both directions. This can be used to support third-party SSO software (e.g. Siteminder) or to maintain repository session cookies.

Does Federated Query Server work with a proxy server?

Yes, the name or address of the proxy server may be specified in the Federated Query Server configuration file. A list of proxy exceptions (sites that should be connected to directly, rather than through the proxy server) may also be given.

Federated Query Server supports Basic, NTLM, Kerberos and delegated proxy authentication.

Does Federated Query Server support HTTPS requests?

Federated Query Server supports SSL 2.0, SSL 3.0 and TLS 1.0 by making use of the OpenSSL library. It also supports HTTPS via a proxy server by using SSL tunneling, and HTTPS reverse proxy servers.

What are the software requirements of Federated Query Server?

Federated Query Server runs under Windows Server 2003, Windows Server 2003 R2, Windows Server 2008, Windows Server 2008 R2, Solaris 9+ (SPARC), HP-UX 11.23+ (Itanium) and Red Hat Enterprise Linux Server 6+ (x86/x64). It requires a CGI-compatible Web server.

What are the hardware requirements of Federated Query Server?

Federated Query Server requires approximately 64MB free RAM and 24MB free disk space. A fast Internet connection (T1 recommended) will help obtain the best performance from Federated Query Server.