|Sesat > Features > Product Whitepaper|
Do you need help to convince your boss
Sesat - SEsam Search Application Toolkit - is a stack of software applications aiming to help efficient creation of feature-rich, stable and multi-source based search portals and search applications. Based on experience and learnings from the first installations of Sesam.no and Sesam.se, Sesat is now an extensive toolkit with one primary focus: reduce development complexity and dramatically improve quality and delivery time. In fact, Sesat lets customers create fully-functional portals in a few days rather than many months, as proved by the teams behind Sesam.no and Sesam.se, all using Sesat to speed up site and site search development.
In general, writing a search application from scratch given only one or more indexes providing search results as data objects or XML, the following steps are usual:
But, this is not at all flexible or scalable, with code that needs to be rewritten and tailored to any change in the underlying data source models. A much better approach would be to:
With Sesat, you get all of the above by
The Sesam Search Application Toolkit, Sesat, is the heart of the http://sesam.no and http://sesam.se search portals. These portals show the power and flexibility of a platform enabling businesses to present both structured and un-structured information from many sources simultaneously, while honoring the user's search context and choices.
Sesat is built on experience and learnings from the first version of the sesam.no portal, and has grown both in maturity and functionality with the tasks given. It has all the time been an important goal to deliver as much flexibility and functionality to the user as possible, while keeping the "core" application stable. Sesam is today successfully leveraging the flexibility of this architecture, focusing on developing their portal rather than the underlying search technology.
Sesat includes a set of applications working together to provide a feature-rich search middleware. Sesat glues together information from underlying indexes and data sources and passes it on to the presentation layer. The toolkit has excellent support for all recent versions of the FAST search engine, as well as many other data sources. It is easy to create own connectors to underlying data layers and sources.
The traditional way of creating a search portal or a search application, is to develop a full stack of components from the lowest layer comminicating directly with search apis, to the templating layer on top.
In effect, Sesat gives you most of these components, so the developer can focus on being productive with front-end coding.
|Overview of the Sesat Model: The developer focus on functionality rather than middleware and interface coding.|
The main parts of Sesat, organising all libaries, applications and tools, are:
|Client applications reside within the Search Front-end Container, SFC. One Sesat-installation can handle multiple client applicatins, or "vhosts", virtual hosts. For example, one single kernel can control and support sesam.no, sesam.se, vg.sesam.no, svd.sesam.se, sports.svd.sesam.se...|
Sesat implements a search portal middeware. It lives in the layer between the application's presentation templates and one or more data sources. It is flexible enough to handle many different types of sources:
The power of Sesat is its ability to search in multiple indexes at the same time, and then create a java object containing all results. This object can then be accessed either via templates (eg Velocity or JSP), or by Ajax-calls. Sesat understands user sessions, and therefore the object can be kept ready for subsequent calls and transactions.
Sesat is controlled by configuration. Some of this configuration is done by means of administrative tools - Sesat ADM. The administrative tools are implemented to run on JBoss Portals, a portal framework compliant to JSR-168, running on JBoss application servers. This framework is also home of several administrative plugins, like personalisation and statistics.
|On the left-hand side, in the Search Front-end Container, the client defines his templates. These are being run by the Sesat Kernel upon user requests. Also, the client provides information about which indexes is to be used, which analysis rules apply etc. In addition, most of the layers in the Kernel define a Service Provider Interface, SPI. The user can implement extensions to functions within the Kernel, ensuring flexibility.|
A full Product Description also exists.
Sesat Kernel runs as a web application within a web-server. The Sesat Kernel, the core to the search front, and the generic.sesam skin (SFC) serves as a search engine's controlling and presentation framework.
Currently, Tomcat and JBoss is supported. All access to a search application or portal, is managed by this part of Sesat. The Kernel is configured to work with a number of Virtual Hosts, vhosts. When starting the service, the Kernel loads information about all active vhosts. Each of these may have their own templates, indexes and configuration. Client code for any given vhost is initialised and loaded at this stage.
The SFC is where client applications run. They are deployed and directed by the Sesat Kernel.
Applications run within the same JVM as the Kernel. They are all deployed as .war-files to the same directory, by default.
The appplication names are the same as the vhost they implement.
Most Kernel functionality is accessible from WebServices and the XML-interface. This is important functionality external applications that needs to communicate with Sesat. For example, mobile applications have been developed to perform searches using the XML API.
Please note that XML-RPC and Web Services to Sesat are not yet implemented - We had great plans for XML-RPC and/or a Web Service API but us, and all our partners, always come back to using RSS (or custom XML) feeds. That's because at the moment the API, like most search APIs, is all RESTful so URL requests and XML (or serialised datamodel) responses gives you the full API.
(source code to ADM not yet released)
The Sesat admin is a collection of applications and tools that run in a JBoss Portals server. The applications interact with various parts of Sesat, for example the Statistics engine.
Sesat Admin consists of portlets and applications running on JBoss Portals (server), which is a standards-compliant (JSR-168) portal. As of version 1.0 of Sesat, the following appliacions are available:
You may use Sesat under the Affero GPLv3 license. This is a Free Software and Open Source license. All your source code including that belonging to derivative works must be made freely available to any user that interacts with the software locally or via any network (eg the internet) unless you choose to bundle it as "aggregated work".