Web Scale Discovery Services: Google like search experience

Web Scale Discovery Service

Google like Search Experience

Nikesh Narayanan M.Tech, MLIS, M.com

Discovery Service Engineer

EBSCO Information [email protected]

Important point #1

0% of users start their research on your

library’s website

(95% close to reality )

Important point #2

90% of users Users prefer Google as their first

search preference

Why users move away from Library ?

• library lacks a research tool that could fully search all the subscribed resources of the Library

• Library website usually present too many points of entry; A variety of different databases make it confusing to figure out where to go.

• Users wanted to start searching immediately from a single entry point

– Books: Library OPAC (ILS

module)

– Articles from Individual e-

Journals

– Various e-Book collections

– Different e-journal publisher portal

– Aggregated : Full text and

Bibliographic Databases

– Databases that are not Subject

Indexes (WoS, Scopus etc.)

– Local Digital Collections (IRs)

Users’ Dilemma

Why users move away from LibraryDisjoint Sources of Information

Where to Start?

Charleston 11-07

Where

should I

begin?

Dissatisfied users

Under utilized resources

Loss of money time and efforts

Unable to provide relevant information to users at right

time

Inefficient mark on Library system ?

Ultimate loss to institution & nation

Adverse consequences

Users Go to Google

But fail to get their relevant and authentic information

What is the solution ?

A Simple Vision

A single point entry to all the content and services offered by

the library…..

Search e-Journals, books, catalogue, IR’s etc.

….as well as providing an enjoyable search experience

Search

• Federated Search Software

• Single Search box

Web scale Discovery is beyond …..

Federated Search Software

Search:

Search Results

Real-time query and responses

Search:

Search Results

Pre-built harvesting and indexing

Meta Search Index

Local Index

OPAC

Digital Collections / IRs

Discovery Services

Discovery Service Federated Search Engines

Search is very fast as retrieval is done in pre-harvested index

Slow (longer time for search completion) –Federated searching is performing the meta search on the fly from different resources

Standardized unified Index Many indexes: Individual indexes and different database structures of various publishers makes it difficult for metadata retrieval.

Robust Relevancy Ranking as retrieval is from Unified index

Testing has proven that relevancy ranking is a major issue in retrieval of quality data.

Metadata Enhancement is possible Metadata enhancement is not possible

Comprehensive results Shallow results -and eventually users will miss much relevant content.

Performance quality is very high Many times important information from relevant resources are missed out due to connection error.

Subject filtering of result set is based on the segmentation ( pre-defined controlledvocabularies)

Subject filtering is through automatic clustering and noise of non- standardized vocabularies may mislead the users.

Discovery Vs Federated Search Engine

Web Scale Discovery Service

Simple yet Powerful solution

As Simple as Google / Google Scholar

Much higher in terms of functionalities

Google like simple search box on Library website

Web Scale Discovery Service

1. Possible to search / filter to Library’s

subscribed resources only.

2. Possible to limit search to peer

reviewed articles

3. Integration of Library Catalogue and

Institutional Repositories

1. Not available

2. Peer Reviewed filtering is not

available

3. Catalogue and IR integration is not

available

Discovery functionalities are much higher than Google Scholar

Web Scale Discovery Service

• More No. of facets • Less facets for filtering

results

Discovery functionalities are much higher than Google Scholar

Web Scale Discovery Service

5. Integration of subject Indexes is

possible through platform blending

6. Integration of SCOPUS and Web of

Science Possible

5. Not available

6. Not available

Discovery functionalities are much higher than Google Scholar

Web Scale Discovery Service

A to Z Directory of Library

Resources

Not available

Discovery functionalities are much higher than Google Scholar

Web Scale Discovery Service

Anytime anywhere with full

text access through

Integration with Remote

access technologies

( eg: Ezproxy, Athens etc. )

Not available

Discovery functionalities are much higher than Google Scholar

Advanced WSD Tools : features

• Integrate of most of the resources

• Non covered resources are integrated through

federated connectors (not possible in Google

Scholar)

• Integration of Subject Indexes through Platfrom

blending

• A to Z directory of Library Resources

• Customization

• Catalog Integration

• IR integration

• Widgets

Main Components

Content (Knowledge Base)

Technology

Content (Knowledge Base)

– Journals

– Books

– Databases

– Aggregators content

– Open source Materials

– News papers, indexes etc.

– Library Catalogue

– Institutional Repositories

Base Index

Local Index

Technology

Harvester (OAI/PMH etc.)

Automated transfer routines, load tables

Metadata mapping

Platform blending

Indexing technology

De-duplication Algorithms

Link Resolvers

Relevancy Algorithms

Interface technologies

Institution Major results

Grand Valley State University Increased use of full text download

North Carolina State University Increased user satisfaction and usage

Oregon State University (worldcat local) Effective search which resulted in increased usage

University of Texas (Summon) Significant Increase in the use of electronic resources

Illinois University (EBSCO Discovery Service) significant increase in database usage.

Comparative Study : Bucknell University and Illinois Wesleyan University – comparison of Discovery services & Google Scholar

EBSCO Discovery Service outperformed other discovery systems & Google Schloar

Discovery Service brings back users to

Libraries : Studies show the Evidence

Mobile Access

Responsive Design for Mobile

phone & Tablet device users

Auto-detection of mobile &

smartphone users...Skinning/Branding…Facets…Usage Reporting

Ideal for Tablets

Enjoy Discovery !

Thanks!