Taxonomies are Dead Long Live Taxonomies Enterprise Search

Taxonomies are Dead, Long Live Taxonomies Enterprise Search Summit May 18, 2005 Sharon Flank, Ph. D. Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC.

Introduction w What makes you think you need a taxonomy? w Goals w Assumptions Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 2

What’s a Taxonomy For? w Organizes corporate information hierarchically, so everyone can find it Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 3

Weaknesses: Individuality w My taxonomy isn’t your taxonomy n How am I to navigate if I don’t know what you were thinking? l How well organized are your PC folders? w How will a corporate taxonomy fit my email? Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 4

Weaknesses: Global View w Is an SUV a truck? w Is a Mini a sports car? w Tomahawk n n Cruise missile Native American weapon Motorcycle Town in Wisconsin w Flickr: everybody helps (!) Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 5

Weaknesses: Time and Mgmt w Taxonomies take time n n Meetings on what goes where, naming Changing world changing taxonomy Anthrax: Disease or WMD? l East Germany, Yugoslavia; Moldova, Latvia l Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 6

Weaknesses: Scale-up w 7 layers x 7 layers… n And then it gets complicated w Automated taxonomy generation n But you still need to edit, disseminate Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 7

Taxonomy vs. Search w Business rules n Enforce a hierarchical view of your business via a taxonomy l Ford’s Volvo and Jaguar divisions w Granularity n All your special nicknames, divisions, abbreviations l SARS, HSARPA Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 8

ROI w Your goal isn’t categorization, or search, or even finding things w You want to save time and make money w Tie the taxonomy to a specific business goal, and measure Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 9

Optimize w Multiple access points n Vacation pix, family pix, best pix? w Multiple modalities (a taxonomy and free search) w Phrases n Venetian blind vs. blind Venetian w Leverage existing resources (taxonomywarehouse. com, findability. org) w Search training Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 10

The Future of Finding w Easy for me to use n n Flexible vocabulary Intuitive or common interface l n Not Boolean, not Google, just English Extensible data format No recoding necessary l Can add metadata l n Anticipates 22 nd century, not just 2010 Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 11

Optimizing w Bad n n n w Good Taxonomy meetings Proprietary formats Guess-my-term Untrained searchers Word salad Limited fields n n n Standard elements Open, upgradeable Synonyms 2+ word searches Syntactic precision Metadata options Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 12

Key Points w Use taxonomies thoughtfully n n n Maintenance Growth Can’t users just search? w Use business goals to determine choice of approach(es) w Design to allow incremental improvement as technology matures Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 13

Thank you Sharon Flank, Ph. D. sflank@datastrategyconsulting. com http: //www. datastrategyconsulting. com Copyright 2004 -2005 Sharon Flank. Data. Strategy LLC. 14