Feature Description





Introduction

ATG Search provides a best-of-breed search engine that integrates
with your ATG Commerce, Service and Portal solutions for quick and
easy deployment. It provides users with powerful tools to search for
information across the enterprise ensuring that they find the product,
service or other content they are looking for, for increased sales,
better service, and better productivity.


Moreover, ATG Search allows you to take advantage of ATGs powerful repository
structure for integrated content retrieval and knowledge management. Specific
search capabilities include:


Two Level Indexing: indexing of content from multiple data sources (including
ATG Repositories, ODBC & JDBC databases, other structured data sources and
unstructured data sources) and of multiple file types (including but not limited
to MS Office, MS Exchange, PDF etc.).

Structured and Unstructured Information retrieval: question ranked results of
content by contextual relevance for higher precision through our proprietary
algorithms, using natural language queries and including as optional the use
of standard Boolean operators.

Natural Language Processing: including orthographic, morphological, lexical,
syntactic, semantic, discourse and pragmatic processes.

Multi-Language Search: ATG Search is capable of searching across multiple
languages simultaneously. Questions in one language (target) can be answered
by content in a second language (source) if content in the target language is
unavailable. Allowing for immediate repurposing of content in a global support
scenario.

Categorization Technology: topics are an integral part of enhancing search
which include cross cutting taxonomies of topics, topic assignment including
rules based topic creation for supporting automatic topic features.

Faceted Search; facets are focused on a specific, limited slice of the information
space; a facet hierarchy can be much smaller and flatter than traditional tree
structures. Even with multiple facet hierarchies there will be relatively few
managed vocabulary terms. Allowing for a large number of content search
combinations from a manageable vocabulary.

Question Gap Analysis: reports that deliver the most frequently asked
questions, questions unanswered by your content and reports assisting in
identifying areas for new content development.

Customer satisfaction feedback loop: tracks and reports on customer metrics

Integration with ATG Knowledge Center, Commerce, Portal and Forum utilizing
the ATG application framework. ATG White Paper 2

Feature Description







What customers have implemented ATG Search?
ATG Search is a mature product with a new name and as expected enjoys a broad
acceptance as the search application in enterprise deployments. ATG Search
formerly Enterprise Search was renamed and launched in Q1 of 2005. ATG Search is
based on leading information management technology developed in-house and
cited by the Seybold group as number one in search result quality compared to
seven competing vendors in the category. At its five-year anniversary, ATG Search
has now served over 100 million answers to the customers of ATG clients including;
Cingular/AT&T Wireless, Symantec, Altera, EMC, and Boeing to name only a few.


How does ATG Search manage security?
ATG Search is designed to work within the security needs of your enterprise to
ensure that users cannot view documents or portions of documents to which they
should not have access. ATG Search integrates with Active Directory and LDAP
systems typically. The control of content is defined through access to Topic areas
based on affiliation with User Groups and User Roles as defined in the Search
management console.

Can ATG Search perform federated search?
ATG Search enables companies to index multiple data sources and see a common
set of ranked results, ranked in order of relevance, regardless of data source what
is often called in the industry a federated search. ATG Search can index data
stored within ATG repositories, in file systems, in structured databases, and on
HTTP servers across the Web. ATG Search can also be supplemented with
additional custom connectors to access information in popular enterprise systems,
including content management systems in Customer Relationship Management
systems and in Enterprise Resource Planning systems for example. Connectors also
allow for content harvesting from proprietary systems and databases native to
your organization.






ATG White Paper 3

Feature Description




How does ATG Search integrate with ATG Portal?
ATG Search includes a portlet for the ATG Portal that allows users to search data
accessible via the ATG Portal and provides configuration options to turn search into
a powerful information management tool. The search interface presents users
with a simple field for natural language search terms or allows them to access an
advanced search interface.
The search portlet displays results in ranked order of relevance. The number of
results or information about results can be configured by the administrator, the
community administrator or the end user. Users can save search terms or can
refine their search, selecting a single result as a key concept for a new search.



How does ATG Search integrate with ATG Commerce?
ATG Search is configured to search any element in the ATG repository including
ATG Commerce content and catalog items. It use will be both for end users of ATG
customer commerce sites as well as merchandisers and internal agents needing
rapid access to specific product and service information. Faceted Search capabilities
will be extended to provide the basis for a broad collection of guided selling
features in the commerce area. In a future release, ATG Search will be integrated
into ATGs commerce reference applications constituting an out-of-the-box and
best-practice implementation of a Search user interface for Commerce sites.

How is search managed?
ATG Search relies on the administration tools provided in the ATG Search
Management Console. This tool allows the administrator to define the data
sources to be indexed, the frequency of indexing and to monitor the performance
of the search servers. Additional content administration specific for an application
e.g. Commerce or Knowledge Center, will be handled within the specific
applications management tool.




ATG White Paper 4

Feature Description





How Does ATG Search Work?


ATGs proprietary technology provides the ability to index and access every
kind of online content -- unstructured content, structured databases, and
proprietary semi-structured information.


ATG search is unique among commercial search engines in that it provides
an index at both the "document" level and at the "statement" level.
Unlike traditional search engines which are document-centric, ATG's
search approach retrieves specific statements within a document,
so the customer does not have to read through many pages of text to see
if the answer is there. The ATG approach to search provides a unified
way to return targeted answers from both structured and unstructured
content. Other providers depend on using NLP to match manually pre-defined
answers, which require teams of human editors in order to build
templates of known questions. ATG Search finds the matching statements
without human intervention. Like many competitors, ATG technology has the
ability to interpret word senses and concepts based on context. Because of
ambiguity, words often match in other engines even when the document that
matches has nothing to do with the query. Although the intent of each question is
quite obvious to a person, it is not obvious to a computer program, and this can be
a tremendous source of error and unsatisfactory response in this type of system.
ATG Search uses our own proprietary language analysis to find the words and
phrases in both the content and the query to provide a good match. Words are
analyzed both "inflectionally" ("broken" is related to "break") and "derivationally"
("install" is related to installation"), in order to provide effective
answers such as "How do I install a new driver" bringing up a section
headed "Installation of new drivers."






















ATG White Paper 5

Feature Description

Understanding the key underpinnings of ATG Search is required in order to
effectively use it. The following section is intended for technical project managers
or leaders evaluating or using ATG Search.

In general, search is the retrieval of information from a repository using an input
query. Commonly, the information is textual, the repository is a collection of
documents, and the queries are words or phrases entered by the user. The results
of search are typically the most relevant documents to the query, plus some
indication of why the document was retrieved. In order for search to be efficient,
the document collection is indexed by its terms in a secondary storage component,
typically called the
index. The following diagram illustrates these basic
components of search: