Friday Mar 30, 2012

Curious About Oracle's BI and Analytics Strategy?

Normally we use this blog space for discussing our business intelligence and analytic efforts along with our views and perspective on this very fast growing marketplace. However, I can't resist mentioning that we are having a great webcast coming up next week, so please do join Oracle's Mark Hurd and Balaji Yelamanchili as they unveil the latest advances in Oracle's strategy for placing analytics into the hands of every one of your decision makers-so that they can see more, think smarter, and act faster. Register now at http://bit.ly/HpAOJk for the Webcast and Live Chat: Wednesday, April 4, 2012 at 9 a.m. PT, 12 p.m. ET, 10 a.m. GMT.  You don't want to miss this event and thank you very much. 

Consumer Oriented Search In Oracle Endeca Information Discovery - Part 2

As discussed in my last blog posting on this topic, Information Discovery, a core capability of the Oracle Endeca Information Discovery solution enables businesses to search, discover and navigate through a wide variety of big data including structured, unstructured and semi-structured data. With search as a core advanced capabilities of our product it is important to understand some of the key differences and capabilities in the underlying data store of Oracle Endeca Information Discovery and that is our Endeca Server. In the last post on this subject, we talked about Exploratory Search capabilities along with support for cascading relevance. Additional search capabilities in the Endeca Server, which differentiate from simple keyword based "search boxes" in other Information Discovery products also include:

The Endeca Server Supports Set Search. 
The Endeca Server is organized around set retrieval, which means that it looks at groups of results (all the documents that match a search), as well as the relationship of each individual result to the set. Other approaches only compute the relevance of a document by comparing the document to the search query – not by comparing the document to all the others. For example, a search for “U.S.” in another approach might match to the title of a document and get a high ranking. But what if it were a collection of government documents in which “U.S.” appeared in many titles, making that clue less meaningful? A set analysis would reveal this and be used to adjust relevance accordingly.

The Endeca Server Supports Second-Order Relvance.
Unlike simple search interfaces in traditional BI tools, which provide limited relevance ranking, such as a list of results based on key word matching, Endeca enables users to determine the most salient terms to divide up the result. Determining this second-order relevance is the key to providing effective guidance.

Support for Queries and Filters.
Search is the most common query type, but hardly complete, and users need to express a wide range of queries. Oracle Endeca Information Discovery also includes navigation, interactive visualizations, analytics, range filters, geospatial filters, and other query types that are more commonly associated with BI tools. Unlike other approaches, these queries operate across structured, semi-structured and unstructured content stored in the Endeca Server. Furthermore, this set is easily extensible because the core engine allows for pluggable features to be added.
Like a search engine, queries are answered with a results list, ranked to put the most likely matches first. Unlike “black box” relevance solutions, which generalize one strategy for everyone, we believe that optimal relevance strategies vary across domains. Therefore, it provides line-of-business owners with a set of relevance modules that let them tune the best results based on their content.
The Endeca Server query result sets are summarized, which gives users guidance on how to refine and explore further. Summaries include Guided Navigation® (a form of faceted search), maps, charts, graphs, tag clouds, concept clusters, and clarification dialogs. Users don’t explicitly ask for these summaries; Oracle Endeca Information Discovery analytic applications provide the right ones, based on configurable controls and rules. For example, the analytic application might guide a procurement agent filtering for in-stock parts by visualizing the results on a map and calculating their average fulfillment time. Furthermore, the user can interact with summaries and filters without resorting to writing complex SQL queries. The user can simply just click to add filters. Within Oracle Endeca Information Discovery, all parts of the summaries are clickable and searchable.

We are living in a search driven society where business users really seem to enjoy entering information into a search box. We do this everyday as consumers and therefore, we have gotten used to looking for that box. However, the key to getting the right results is to guide that user in a way that provides additional Discovery, beyond what they may have anticipated. This is why these important and advanced features of search inside the Endeca Server have been so important. They have helped to guide our great customers to success. 

Friday Mar 16, 2012

Consumer Oriented Search In Oracle Endeca Information Discovery – Part 1

Information Discovery, a core capability of Oracle Endeca Information Discovery, enables business users to rapidly search, discover and navigate through a wide variety of big data including structured, unstructured and semi-structured data. One of the key capabilities, among many, that differentiate our solution from others in the Information Discovery market is our deep support for search across this growing amount of varied big data. Our method and approach is very different than classic simple keyword search that is found in may information discovery solutions. In this first part of a series on the topic of search, I will walk you through many of the key capabilities that go beyond the simple search box that you might experience in products where search was clearly an afterthought or attempt to catch up to our core capabilities in this area. Lets explore.


The core data management solution of Oracle Endeca Information Discovery is the Endeca Server, a hybrid search-analytical database that his highly scalable and column-oriented in nature. We will talk in more technical detail about the capabilities of the Endeca Server in future blog posts as this post is intended to give you a feel for the deep search capabilities that are an integral part of the Endeca Server.


The Endeca Server provides best-of-breed search features aw well as a new class of features that are the first to be designed around the requirement to bridge structured, semi-structured and unstructured big data. Some of the key features of search include type a heads, automatic alphanumeric spell corrections, positional search, Booleans, wildcarding, natural language, and category search and query classification dialogs. This is just a subset of the advanced search capabilities found in Oracle Endeca Information Discovery.


Search is an important feature that makes it possible for business users to explore on the diverse data sets the Endeca Server can hold at any one time. The search capabilities in the Endeca server differ from other Information Discovery products with simple “search boxes” in the following ways:

The Endeca Server Supports Exploratory Search

Enterprise data frequently requires the user to explore content through an ad hoc dialog, with guidance that helps them succeed. This has implications for how to design search features. Traditional search doesn’t assume a dialog, and so it uses relevance ranking to get its best guess to the top of the results list. It calculates many relevance factors for each query, like word frequency, distance, and meaning, and then reduces those many factors to a single score based on a proprietary “black box” formula. But how can a business users, searching, act on the information that the document is say only 38.1% relevant? In contrast, exploratory search gives users the opportunity to clarify what is relevant to them through refinements and summaries. This approach has received consumer endorsement through popular ecommerce sites where guided navigation across a broad range of products has helped consumers better discover choices that meet their, sometimes undetermined requirements. This same model exists in Oracle Endeca Information Discovery. In fact, the Endeca Server powers many of the most popular e-commerce sites in the world.

The Endeca Server Supports Cascading Relevance.

Traditional approaches of search reduce many relevance weights to a single score. This means that if a result with a good title match gets a similar score to one with an exact phrase match, they’ll appear next to each other in a list. But a user can’t deduce from their score why each got it’s ranking, even though that information could be valuable. Oracle Endeca Information Discovery takes a different approach. The Endeca Server stratifies results by a primary relevance strategy, and then breaks ties within a strata by ordering them with a secondary strategy, and so on. Application managers get the explicit means to compose these strategies based on their knowledge of their own domain. This approach gives both business users and managers a deterministic way to set and understand relevance.


Now that you have an understanding of two of the core search capabilities in Oracle Endeca Information Discovery, our next blog post on this topic will discuss more advanced features including set search, second-order relevance as well as an understanding of faceted search mechanisms that include queries and filters.


 

About

We're taking the pulse of the Business Intelligence and Analytics market based on our insights and our experiences with colleagues, customers,and partners.

Search

Categories
Archives
« March 2012 »
SunMonTueWedThuFriSat
    
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
17
18
19
20
21
22
23
24
25
26
27
28
29
31
       
Today