Finding Answers: Going Beyond Search

Finding the right answers, at the right time, and with the right evidence is the ultimate goal of any user’s search. However, results from a keyword search in today’s enterprise systems can be so large that finding relevant information quickly may be difficult, and finding answers to a specific problem might be next to impossible without a significant investment of time sifting through the results. This is only amplified by the proliferation of content across the Web, in enterprise systems and across other information sources (subscription sites, email, internal documents, etc.). In many cases, users must search within each of these sources individually, understand what is contained in each document and manually determine if the results are a good match. The problem facing users today is that they typically do not want more search results; they want more relevant ones.

With only a text box and search button, users have no way of specifying what they want beyond a few keywords. This is only one of many problems with current search technologies:

• If users are not aware of implicit relationships, they may not include needed keywords to get the best results possible. A product manager researching the market for iPods may not know that mobile phones with MP3 capabilities are a critical adjacent market;
• When users provide a keyword, they are not able to limit their results based on a specific meaning of the keyword. When searching for "sap," users get results for both the software company and maple syrup raw ingredients; and
• Today’s search solutions are not able to look across all information sources. Content inaccessible by the search solution won’t be revealed in the search results and is effectively inaccessible to users.

Understanding Context through Text Analysis
Users need a solution, built on their current search infrastructure, that understands the contents and relationships contained within search results. By providing this information, users can immediately see common themes, uncover hidden relationships and logically navigate through a list of results to quickly find the answers they need. Text analysis solutions leverage existing information infrastructures to linguistically analyze the actual structure of sentences, paragraphs and documents to extract entities (people, places, things, events, etc.) and their relationships to each other. This essentially automates what users do every time they search—they read articles, identify important facts and then make a determination on whether each document is relevant. By doing this iteratively and refining search terms each time to get better results, the user eventually understands more about the subject, and hopefully finds the answer to their original question.

If a search is done on "MegaCo" and a document shows it was purchased by SuperCo, we know these entities are related throughout the entire result set. If in another document we know that John is the CEO of MegaCo, and Paul is the CEO of SuperCo, we now know these two people are related. Following this chain of relationships enables the user to find answers to their questions even if they started with no background in the topic. Automating this process with text analysis practically eliminates the time it takes users to gain the added insight uncovered within these links.

Federating Search
Since the number of relevant information sources is growing at an exponential rate, users desperately need a solution that can leverage the discrete search capabilities from each of these sources to aggregate, understand and consolidate results into a single view. By federating search results from multiple systems, users are able to access an unlimited number of sources regardless of vendor or data type. Typically, each source system does the best job of indexing its own content, so it makes sense to leverage these search capabilities rather than re-crawling the repositories to create a unified index. However since each system’s ranking algorithms will be different, it is imperative for federation solutions to understand the result contents beyond just keyword analysis. By doing so, a single view can be created that combines results intelligently and maintains an accurate relevancy ranking across all information sources

Text Analysis + Federated Search = Finding Answers
SAP BusinessObjects Intelligent Search is a powerful solution that enables users to find the most complete and relevant answers to any problem. It is not a search engine itself, but increases the value of your existing information infrastructure by adding text analysis and federated search capabilities in a single integrated solution. Intelligent Search goes beyond search by automatically accessing any number of search sources, understanding the content of each result, identifying relationships across results and providing its recommendation for the most relevant content based on the context supplied by the user.

Since the content of each result is linguistically analyzed, a common ranking algorithm can be applied across all results to provide the user with a true single, unified, relevance-ranked list. Rather than going through hundreds of documents searching for a needle in a haystack, the user can navigate through areas of interest and easily focus on only the most relevant articles that have been pre-read and pre-filtered for them.

Intelligent Search integrates with all existing search engines, portals, content management systems and business applications to help business users and analysts alike to find answers trapped within Internet, enterprise and subscription sources and make better decisions. You can find more information at www.businessobjects.com/product/catalog/intelligent_search

---

SAP software helps organizations gain better insight into their businesses, improving decision-making and enterprise performance. Our mission is to transform the way the world works. We are not just a business intelligence company. We are in the business of helping companies become more intelligent.