Walk Before You Run: Prerequisites to Linked Data

Post on 21-Aug-2015

18 views 0 download

Tags:

transcript

Walk Before You Run

Prerequisites to Linked DataKenning Arlitsch

Dean of the Library@kenning_msu

Linked Data applications will not matter if search engines can’t find library websites and repositories, crawl them, and understand the metadata provided.

First, Take Care of Basics

AgendaTraditional SEO (Search Engine Optimization)– Hardware, software, websites, metadata

Semantic Web Optimization– Semantic Identity– Schema.org Project at MSU• Using a vocabulary understood by search engines• Improve machine comprehension

Funded Research• 2011-2014

– “Getting Found: Search Engine Optimization for Digital Repositories”• 2014-2017

– “Measuring Up: Assessing Accuracy of Reported Use and Impact of Digital Repositories

– Partners• OCLC Research• Association of Research Libraries• University of New Mexico

SEARCH ENGINE OPTIMIZATIONPart 1 of 3

SEO Building Blocks• Priority 1 – Increase Reach– Get objects indexed by search engines

• Priority 2 – Increase Visibility in SERP– Provide robust descriptive content

• Priority 3 – Get Relevant– Increase click-through rates (CTR)

Why it Matters

DeRosa, Cathy, et al. “Perceptions of Libraries, 2010: Context and Community: A

Report to the OCLC Membership”, OCLC, 2010.

Where College Students Begin Research - 2010

* http://www.comscore.com/Insights/Market-Rankings/comScore-Releases-November-2014-U.S.-Desktop-Search-Engine-Rankings

Americans submit 18 billion search queries to search engines each month*• 12 billion to Google sites (67%)• 3.5 billion to Microsoft sites (19%)• 1.8 billion to Yahoo! Sites (10%)

How much of that traffic is directed to our libraries?

Need more reasons?

Our Research Inspiration• Decade building digital libraries - Univ of Utah– Mountain West Digital Library– Utah Digital Newspapers– Western Waters Digital Library– Western Soundscape Archive

• Were they being used…?

Uh, not really…

• 2010 situation at Utah– 12% of digital collections indexed by Google– 0.5% of Utah’s IR scholarly papers accessible via

Google Scholar

Basic SEO began producing significant increases in the average number of page views per day…

Avg. Page Views / Day content.lib.utah.edu

Basic SEO improved Utah’s collection accessibility in Google…

Average

0% 25% 50% 75% 100%

92%

79%

51%

12%

07/05/10 04/04/11 11/30/11 12/05/13

Google Index Ratio - All Collections*

* Google Index Ratio = URLs submitted / URLs Indexed by Google** ~150 collections containing ~170,00 URLs (07/2010) and ~170 collections containing ~282,000 URLs (12/2013)

…resulting in more referrals and visitors

12 week comparison 2010 vs. 2012

Technical Barriers to SE Crawlers• Website Design

– Graphics– Confusing site hierarchies and paths

• Slow servers• CMS often lack canonical links• Metadata

– Schema not understood by SE– Not unique– Inconsistent/inaccurate

Nearly 100% USpace IR content indexed in Google

Google Index Ratio

Board of Regents

UScholar Works

ETD 2

ETD 1

0% 25% 50% 75% 100%

97%

98%

98%

97%

47%

51%

68%

69%

4%

23%

0%

12%07/05/1011/19/1010/16/11

Google Scholar Index Ratio

~0%*October 16, 2011 Weighted Average Google Index Ratio = 97.82% (10,306/10,536).

Challenge is presenting structured data SE’s can identify, parse and digest

Wolfinger, N. H., & McKeever, M. (2006, July). Thanks for nothing: changes in income and labor force participation for never-married mothers since 1982. In 101st American Sociological Association (ASA) Annual Meeting; 2006 Aug 11-14; Montreal, Canada (No. 2006-07-04, pp. 1-42). Institute of Public & International Affairs (IPIA), University of Utah.

Human Readable

Google ScholarUnderstandable

Google Scholar can read and understand!Google Scholar

SEO Organizational/Cultural Themes• Traditional SEO is an afterthought• Librarians think too small re potential traffic• Organizational communication is poor• Analytics are usually poorly implemented• Vendors are slow to catch on to SEO problems– Because we don’t demand it

Recommended SEO Process1. Institutionalize SEO

● Strategic Planning● Accurate Measurement Tools

2. Traditional SEO● Get Indexed = Index Ratio● Get Visible = Search Engine Results Page (SERP)

Advanced SEO Programs3. Semantic SEO

● Get Relevant = Click Through Ratios (CTR)● Semantic Identity● Schema.org for Libraries● Linked Open Data (LOD)

4. Social Media Optimization● Faculty Outreach

SEMANTIC IDENTITY

For Accurate Representation on the Web

12/09/2014

Current SituationAcademic organizations are poorly represented on the Semantic Web…

…because search engines don’t understand them…

…because we don’t maintain the data sources search engines trust.

Affects reputation of the entire academic institution

Colleges

Departments Centers Institutes

Institutional reputation

Researcher collaboration/employment

Research funding

University rankings

Student enrollment

Manage Risk

Google’s Knowledge Graph

The Web is moving from “strings” to “things”

“A knowledge base … to enhance search results with semantic-search information gathered from a wide variety of sources”

Source: Wikipedia

Knowledge Graph Products• Answer Box– Facts about concepts

• Carousel– Group of instances that comprise a concept

• Knowledge Card– Displays information about organizations and

people

Lack of a Knowledge Card in search results is indicative of a larger problem…

…and as a result Google is unlikely to connect users with the organization’s website

…it means Google doesn’t understand that the organization exists or what its business is…

Survey of ARL Libraries• n=125• Searched by name listed in ARL directory• Knowledge Card? Yes/No• Robustness scale of 1-5

Survey of ARL LibrariesNo Knowledge Card at all

43Have Knowledge Card

82 -10 incorrect

-29 (robustness of 1)Total = 43

Google’s Perception of MSU Lib - 2012

MSU Library - 2014

Where does Google get its information?

Trusted Sources for Search Engines• No Wikipedia presence? – Organization doesn’t exist as an “entity” or “thing”– It exists as a string of (confusing) text

• Other influences on Google’s Knowledge Graph– FreeBase (phasing out in favor of Wikidata)– Google Places/Google My Business– Google+

Wikipedia - 2012

DBPedia entry - 2012

2014 DBpedia entry

MSU COLLEGES

MSU CENTERS AND INSTITUTES

Summary• Define library organization in Wikipedia– Beware of *pedia culture and process

• Engage with other trusted data sources– Wikidata– Google Places/Google My Business– Google+

• Mark-up metadata with Schema.org

New Knowledge Work for Libraries• Build set of replicable services– Populate and maintain structured data records– Add rich semantic markup to websites

• Communicate– Understand ourselves from stakeholder perspective– Machine-understandable information

SCHEMA.ORG PROJECTPart 3 of 3

57

Schema.org • Common vocabulary for describing things on web • Supported by Bing, Google, Yahoo and Yandex • “On-page markup helps search engines

understand the information on webpages and provide richer results.”

• https://support.google.com/webmasters/answer/1211158?hl=en

Hypothesis• Implementing Schema.org in library websites– Improves machine understanding of content– Improves rich snippets shown in SERP– Increases click-through rates from SERP

• Result– More traffic– More users finding what they’re looking for

Project: A Controlled Experiment by Jason Clark (with Michelle Gollehon)

• Two digital collections• Similar size/content/date range– Photos and historical documents

• 1 optimized with Schema.org (Schultz)• 1 control (Brook)

A Revised Digital Library Architecture• Collection Page (home page)

– arc.lib.montana.edu/schultz-0010/• About Pages (about page, topics page)

– arc.lib.montana.edu/schultz-0010/about.php• Item Pages (individual record page)

– arc.lib.montana.edu/schultz-0010/item/31• Sitemap and rel=canonical work

– arc.lib.montana.edu/schultz-0010/

Results

Semantic Web Team• Kenning Arlitsch, Dean @kenning_msu• Patrick OBrien, Semantic Web Director @sempob• Jeff Mixter, Research Associate, OCLC Research• Jason Clark, Head of Lib Informatics and Computing @jaclark• Scott Young, Digital Initiatives Librarian @hei_scott• Doralyn Rossmann, Head of Coll Development @doralyn• Jean Godby, Senior Research Associate, OCLC Research

Relevant Publications• Arlitsch, Kenning, and Patrick S. OBrien. (2013) Improving the visibility and use of digital repositories through

SEO. Chicago: ALA TechSource. ISBN-13: 978-1-55570-906-8

• Mixter, Jeff, Patrick OBrien and Kenning Arlitsch. “Describing Theses and Dissertations using Schema.org,” Proceedings of the International Conference on Dublin Core and Metadata Applications 2014, Dublin Core Metadata Initiative: 138-146.

• Arlitsch, Kenning. “Being Irrelevant: How Library Data Interchange Standards have kept us off the Internet,” Journal of Library Administration, 54, no. 7 (2014): 609-619.

• Arlitsch, Kenning, Patrick OBrien, Jason A. Clark, Scott W.H. Young and Doralyn Rossmann. “Demonstrating Library Value at Network Scale: Leveraging the Semantic Web with New Knowledge Work,” Journal of Library Administration, 54, no. 5 (2014): 413-425.

• Arlitsch, Kenning, Patrick OBrien, and Brian Rossmann. "Managing Search Engine Optimization: An Introduction for Library Administrators." Journal of Library Administration 53, no. 2-3 (2013): 177-188.

• Arlitsch, Kenning, and Patrick S. O'Brien. "Invisible institutional repositories: Addressing the low indexing ratios of IRs in Google Scholar." Library Hi Tech 30, no. 1 (2012): 60-81.