Oracle Database 11g Search Features

October 13, 2010

I must admit that I don’t follow Oracle’s search and retrieval initiatives with the assiduity of the past. My goodness there is so much activity in the NoSQL world that I find that the new is driving out the old. A reader sent me a link to a useful write up by Richard Foote. “Oracle OpenWorld Day 5 Highlights.” The article includes a useful summary of search features and I wanted to capture his thoughts and urge you to read his original post. On the other hand, you could navigate to www.oracle.com, use the Web site search function, and locate the information yourself. (It did take me some time to find what I sought, but your mileage may vary.)

I learnt to my surprise that there are a quite a number of  new features in relation to Oracle Text in the recent 11.2.0.2 release. New features include Entity Extraction whereby Oracle will automatically find entities in text such as people, cities, phone numbers, etc. a new Name Search facility in which people names with different spelling can more easily be found (such as Stephen and Steven) and a new Resultset Interface capability in which details and data can be nicely summarized. Also mentioned are enhancements in the manner by which frequent and not so frequent accesses to text tokens can be stored and processed. Also had a really interesting look at what new things are being planned, such as automatic partitioning, automatic optimizations of indexes via the use of a staging index, section specific index options, two index levels with better management of common terms in memory, substring index options to name but a few. Looks like there are going to be considerable functional improvements to text indexes on their way soon.

Oracle is a very large outfit, and I must invest some time in figuring out what happened to TripleHop, SES11g, Oracle Text, and the search functions that once were resident within such Oracle acquisitions as PeopleSoft. Soon. Hopefully soon. More rumors about a deal between Oracle and a search vendor. No comments on that, however.

Stephen E Arnold, October 13, 2010

Freebie

Comments

2 Responses to “Oracle Database 11g Search Features”

  1. Avi Rappoport, Search Tools Consulting on October 13th, 2010 11:40 am

    For what it’s worth, PeopleSoft didn’t have much in the way of text search. I was working with them to improve their intranet Verity installation when they were bought.

    Avi

  2. Stefan Buchta, Oracle Search Product Mgmt. on October 13th, 2010 9:47 pm

    Stephen — nice overview of 11g RDBMS new Text features, I’m glad to see it on your page. A better title for this news item might be “Oracle 11g Database Text Features”. Yes, I realize Search was in the original title of the OpenWorld talk, but to continue using the “Search” keyword here might be a bit misleading. The new features (ie. entity recognition, power name search) mentioned in the OpenWorld blog belong to Oracle Text, an unstructured text processing engine and a feature of the Oracle database. Text is designed as a ‘platform’ — a power user oriented data retrieval module used to enable unstructured data indexing within database-backed applications (some example customers are companies running very large archives (IronMountain), cataloges (LandsEnd), discovery- and compliance tools used by Discovery service vendors (Epiq)). It assumes that your data is already in the database and you are trying to full text enable it… so it does not contain a crawler or any OOTB GUIs. Instead, Oracle offers Secure Enterprise Search (SES), a J2EE application built on top of Text for customers who want a complete search engine. Not all Text features are exposed through SES today, but customers can expect that the described innovations to become available in future SES 11 releases.

    Oracle PeopleSoft are replacing Verity and working towards embedding SES. Other Oracle ERP & CRM groups like Siebel and EBusiness Suite are already shipping with SES.

    Stefan