E N D
1. NCSU OPAC Search & NavigationPete BellWiLSWorld 2006
2.
Page 2 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com 1911 — New York Public Library
3.
Page 3 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com
4.
Page 4 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com
5.
Page 5 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com 2003 — Bibliotheca Alexandrina, Egypt
6.
Page 6 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com 2004 — Seattle Public Library
7.
Page 7 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com
8.
Page 8 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com 1953, NCSU D. H. Hill Library
9.
Page 9 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Agenda Endeca & Guided Navigation
NCSU’s OPAC
NCSU results
NCSU project details
The future?
10.
Page 10 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Endeca’s mission To help people
find, analyze, and understand
information in ways never
before possible MAIN POINTS
…To help people find, analzye, and understand information in ways never before possible
We do this so that people can make decisions and take actions that they couldn’t before
When users accomplish their goals, organizations fulfill their business objectives
TRANSITION
This mission compels us to create innovative technology. And that innovation drives our growth
MAIN POINTS
…To help people find, analzye, and understand information in ways never before possible
We do this so that people can make decisions and take actions that they couldn’t before
When users accomplish their goals, organizations fulfill their business objectives
TRANSITION
This mission compels us to create innovative technology. And that innovation drives our growth
12.
Page 12 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Elements of Information Access Information Architecture
Backend
User Interface
13.
Page 13 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com But how does it scale?
14.
Page 14 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com But how does it scale?
15.
Page 15 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Elements of Guided Navigation Information Architecture
Faceted hierarchical metadata
Backend
Meta-relational index
User Interface
???
16.
Page 16 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Benefit of post-coordinate, faceted navigation Four independent categories [facets] of 10 nodes each can have the same discriminatory power as one hierarchy of 10,000 nodes.
-Joseph Busch, Taxonomy Strategies
17.
Page 17 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Backend:Meta-Relational Index
18.
Page 18 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Scaling Guided Navigation (log)
19. Guided Navigation in Action
20.
Page 20 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Where is the value of all that investment in cataloging, if the user just types in a search like this?
There should be more than simply the “statistical probability” that they happen to match a cataloging term that adds that book to the list of results when it otherwise would not have been retrieved by matching that word in the books description blurb (estimated by Arlene Taylor, an emeritus professor of Library Science at U Kentucky, at roughly 33% based on her log research…)Where is the value of all that investment in cataloging, if the user just types in a search like this?
There should be more than simply the “statistical probability” that they happen to match a cataloging term that adds that book to the list of results when it otherwise would not have been retrieved by matching that word in the books description blurb (estimated by Arlene Taylor, an emeritus professor of Library Science at U Kentucky, at roughly 33% based on her log research…)
21.
22.
Page 22 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Search brings back the whole tail
23.
24.
26.
27.
28.
29.
30.
31.
Page 31 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com The reference interview
32. Success StoryNorth Carolina State University
33.
Page 33 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Document downloads from Web site search
34.
Page 34 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Document downloads from Web site search
35.
Page 35 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com NCSU
36.
Page 36 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com
37.
Page 37 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Did you know? A typical library of 1MM books spends approximately $20-30MM per year . . .
To purchase them
To catalog them
To maintain/bind them
To shelve them
On systems that manage their circulation
On real estate & building maintenance to house them
On annual purchases to augment collections
How do we get a return on that kind of investment?
38.
Page 38 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com NCSU Objectives Increase usage of university library by students, faculty, and other members of the community
Number of items in circulation at any given time
Increase re-circulation
better return on investment in collection
Increase usage of legacy collection
Most recent always at the top of results (although maybe not most relevant)
Increase % of Successful Searches
Combat perceived lack of coverage
Better control over application, relevancy; more flexibility
Easy of maintenance
39.
Page 39 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com NCSU - Pursuit of Features Endeca
Speed
Relevance Ranking
Faceted Browsing
True Browsing (LC)
Data-driven spell-checking
Automatic stemming
“Did you mean…”
Hierarchical browsing
Browsing in CONTEXT of search
Scale & operational simplicity
Ease of use
Unicorn / Web2
As if…
Last-in / First-out
Authority index links
Query required
Dictionary lookup only
No
No
No
No
No
No
40.
Page 40 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Implementation Resources One java-trained librarian (not a developer) working 30-40 hours per week for 14 weeks
Worked with librarian from Digital Library Initiatives to do wireframing and fleshing out requirements (total of 40-60 hours)
Project Management by Andrew Pace approximately 10 hours per week for 20 weeks
Committee consisting of a cataloger for metadata issues and a reference librarian for interface issues – total of 2-4 hours per week over 12 weeks (combined)
41.
Page 41 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Pre-Endeca Catalog Search Problem: How to provide Endeca keyword searching and Web2 authority searching while keeping the search interface as close to the ‘one box’ approach as possible.Problem: How to provide Endeca keyword searching and Web2 authority searching while keeping the search interface as close to the ‘one box’ approach as possible.
42.
Page 42 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Endeca Catalog Search
43.
Page 43 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com
44.
Page 44 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Some user reaction “This is absolutely the coolest thing I've seen all century.”
Will Owen, Head of Systems (UNC Libraries)
“Also, I'm really digging the new NCSU library catalog. Very nice."
- Educause staff (non-librarian)
“The new Endeca system is incredible. It would be difficult to exaggerate how much better it is than our old online card catalog (and therefore that of most other universities). I've found myself searching the catalog just for fun, whereas before it was a chore to find what I needed.”
- NCSU Undergrad, Statistics
45. Technical Overview
46.
Page 46 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Technical Overview Endeca co-exists with SirsiDynix Unicorn ILS and Web2 online catalog.
Endeca indexes MARC records exported from Unicorn.
Index is refreshed nightly with records added/updated during previous day.
47.
Page 47 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Basic Architecture
48.
Page 48 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Step 1: Data Transformation
49.
Page 49 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Step 2: Data Pipeline Editing
50.
Page 50 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Step 3: Create Dynamic Dimensions
51.
Page 51 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Step 4: Create Edited Dimensions
52.
Page 52 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Step 5: Load Indices and Create UI
53.
Page 53 © Endeca Technologies Inc. All rights reserved.
Proprietary and Confidential
www.endeca.com Thank you! Pete Bell