@WebSciDL PhD Student Project Reviews August 5&6, 2015

Web$Science$and$Digital$Libraries$Research$Group$$

@WebSciDL$

Review$of$Projects$for$$Herbert$Van$de$Sompel,$LANL$

August$5&6,$2015$$

Corren McCoy

Disambiguation of Alumni from Publicly Available Social Media Profiles

Presentation for Herbert Van de Sompel 08/05/2015

Let’s be Social!

Directory Search Name: Michael Nelson College: Old Dominion Degree: Computer Science Year: 1997

Motivation

Maintain relationships with alumni

Interact and re-engage

Pew Research Survey, Sept. 2014 LinkedIn is used by 28% of online adults. 23% are between 18-29*

Twitter is used by 23% of online adults. 37% are between 18-29

*Pew Research Center noted a significant change in this percentage from 2013

Research Goals • Given discrete set of attributes • Leverage public information

• Collect structured/unstructured metadata • Develop a probabilistic matching scheme

• Analyze and discover new profile attributes • Connect the networks

Seminal Works

• Mislove, A., Viswanath, B., Gummadi, K. P., & Druschel, P. (2010, February). You are who you know: inferring user profiles in online social networks. In Proceedings of the third ACM international conference on Web search and data mining (pp. 251-260). ACM.

• Northern, C. T., & Nelson, M. L. (2011). An unsupervised approach to discovering and disambiguating social media profiles. In Proceedings of Mining Data Semantics Workshop.

• Powell, J., Shankar, H., Rodriguez, M., & Van de Sompel, H. (2014). EgoSystem: Where are our Alumni?. Code4Lib Journal, (24).

Our Work is Informed

Attribute inference based on a Facebook crawl of a known friends network with matching to a Student or Alumni Directory. Examination of digital preservation strategies across social media sites using feature data to score and disambiguate the discovered profiles. Aggregation of discovered social and institutional artifacts to a public identity which are linked in a property graph to facilitate searching.

Mislove

Northern

Powell 6

Similarity Metrics

Does it help to know a name?

Census Surnames Social Security Administration

Name Ranking as of 2014 Michael 7 Nelson 40

Michele ----- Weigle 13,604

First names 19,584 Surnames 150,436

Are Vanity Screen Names Re-used?

LinkedIn: michaellloydnelson Twitter: phonedude_mln

Is the Affiliation Repeated?

LinkedIn: Old Dominion University Twitter: Old Dominion University mentioned in bio but could be a false positive

How Far Apart in Space?

LinkedIn: Norfolk, Virginia Area Twitter: Norfolk, VA

Do People Re-use Profile Photos?

TinEye Reverse Image Search

Do Web Links Point to the Same Page?

LinkedIn: http://www.cs.odu.edu/~mln/ http://ws-dl.blogspot.com/ http://f-measure.blogspot.com/ Twitter: cs.odu.edu/~mln/

Community Analysis Surrogate Connections - People Also Viewed

One step from Dr. Nelson One step from Brittany Johnson

Community Analysis Disclosed – (Followers?) and Following

Property Graph Analysis

https://twitter.com/phonedude_mln

https://www.linkedin.com/in/michaellloydnelson

Location

Norfolk, Virginia area Norfolk, VA

Location

Affiliation Value: Old Dominion

Attended

Geo-Location

Attended

Twitter @ODUNow

hasOfficialAccount

Geo-Location

Attended

Twitter @ODUNow

hasOfficialAccount

follows

Example Searches

LinkedIn Candidate Search

• Leverage Google’s advanced search operators to improve precision.

• Trusted information from the Registrar’s Office.

LinkedIn Metadata How Prevalent are Nicknames?

Name Michael Nelson Mike Nelson Mike Nelson

Headline Professor at Old Dominion University Orthotist / Certified Athletic Trainer Driver at Old Dominion Freight Line

Location Norfolk, Virginia Area Providence, Rhode Island Area Phoenix, Arizona

URL https://www.linkedin.com/in/michaellloydnelson

https://www.linkedin.com/in/mikenelson64

https://www.linkedin.com/pub/mike-nelson/6b/50b/879

Profile Photo https://media.licdn.com/mpr/mpr/shrinknp_400_400/p/1/000/019/1d1/39275de.jpg

https://media.licdn.com/mpr/mpr/shrinknp_400_400/p/2/000/02f/11d/3f17849.jpg

Vanity Screen Name michaellloydnelson mikenelson64

Industry Research Hospital & Health Care Transportation/Trucking/Railroad

Websites http://www.cs.odu.edu/~mln/ http://ws-dl.blogspot.com/ http://f-measure.blogspot.com/

----- ----

Affiliation(s) Old Dominion University, 1997-2000 Old Dominion University, 1996-1997 Virginia Polytechnic Institute and State University, 1987-1991

Old Dominion University, 1999-2001 -----

Twitter Candidate Search

Twitter Metadata Given and Nickname Search

User Name Michael L. Nelson Mike Nelson Mike Nelson

Head of @WebSciDL, Computer Science, Old Dominion University; Formerly: @NASA_Langley (1991-2002), @UNCSILS (2000-2001); OAI-PMH OAI-ORE Memento ResourceSync

----- -----

Location Norfolk, VA ----- -----

URL https://twitter.com/phonedude_mln https://twitter.com/mikenelson64

Profile Photo https://pbs.twimg.com/profile_images/959295176/mln-ad-100x130_400x400.jpg

----- -----

Screen Name Phonedude_mln mikenelson64

Industry ----- ----- -----

Websites cs.odu.edu/~mln/ -----

Affiliation(s) Old Dominion University in bio. Following @ODUNow official account

----- -----

Known Issues • Reliability of Name Searches

– Nicknames list from the Northern (2011) study is incomplete. Ignores ethnic given names.

– Given and surname data from US census and SSA must exist at a certain threshold to protect privacy.

– Naïve calculation of name probabilities. Some name combinations do not occur frequently.

• Uncovering social data is difficult – LinkedIn limits use of API to get real connections. – Rate limits on the Twitter API constrain the depth of

the followers/following search.

Known Issues • Each network takes a different approach to

the visibility of metadata – Exploit the structure of LinkedIn – Twitter data is noisy, limited space with no

controlled vocabulary

By: Alexander Nwala August 5, 2015

Progress Report

Presented To: Dr. Herbert Van de Sompel, Dr. Michael Nelson

Progress Report

Outline• Past projects

• Refactoring Hany’s Carbon date • What Did It Look Like? • I Can Haz Memento

• Present research • Exploration of Distributed Information Retrieval

• Problem • Goal • Research paths; possibility contributions

Carbon date• Estimates the creation date of a URI • The current implementation features a:

• Threaded server • Concurrent API requests • Cached responses

• This is achieved by picking the least date from these sources:

• Last modified date • Bitly • Topsy • Backlinks • Archives

Website: http://cd.cs.odu.edu Blog post: http://ws-dl.blogspot.com/2014/11/2014-11-14-carbon-dating-web-version-20.html

What Did It Look Like?

• Tumblr blog which • Uses the Memento framework to poll various public web archives • Creates an animated image for each year that shows the progression of the site

through the years • Everyone is free to nominate web sites to What Did It Look Like? by tweeting:

“#whatdiditlooklike URL”

Website: http://whatdiditlooklike.mementoweb.org/ Blog post: http://ws-dl.blogspot.com/2015/01/2015-02-05-what-did-it-look-like.html

I Can Haz Memento

• Inspired by the “#icanhazpdf” movement and also built upon the Memento framework

• For tweets with links containing “#icanhazmemento” • I Can Haz Memento service replies the tweet with a link pointing to:

Website: https://twitter.com/icanhazmemento/ Blog post: http://ws-dl.blogspot.com/2015/07/2015-07-22-i-can-haz-memento.html

Archived version of the page closest to the time of the tweet

Progress Report

Outline• Past projects

• Refactoring Hany’s Carbon date • What Did It Look Like? • I Can Haz Memento

• Present research • Exploration of Distributed Information Retrieval

• Problem • Goal • Research paths; possibility contributions

Problem :: Undiscoverable resources are not included in SERPs

• SERP does not have intended resource: “A kinetic theory for age-structured stochastic birth-death processes”

• But resource is available in a special collection (arXiv.org)

Case 1, SERP for Query: “stochastic birth-death processes”

Google Search

arXiv.org Search

Problem :: Information not discoverable from Google do not exist to many web users

• 1st page of SERP does not have intended resource: “EPIDEMIOLOGY THROUGH CELLULAR…”

Case 2, SERP for Query: “influenza indonesia”

Google Search

arXiv.org Search

Relevant resource on 7th page

Relevant resource on 1st page

Problem :: Inconsistent views between SERP and special collections

Problem :: When to stop?

• A user potentially misses relevant information because it is NOT presented with search results OR presented too far (e.g. last 7th page)

• In other words, if relevant content is not presented in the first n pages (e.g. n < 3), it does not exist

Goal :: Present resources from multiple unindexed sources with Google SERP

• This can be achieved through middleware such as a browser plugin

10 more relevant resources1.

2. Click

Relevant resource on 1st page

Exploration of DIR :: Problem summary and Goal

• Problem • Inconsistent views between SERP and special collections

leads to absence of relevant resources in SERPs (Case 1)

• If relevant content is not presented in the first n pages (e.g. n < 3), it does not exist (Case 2)

• Goal • Present resources from multiple unindexed sources with

Google SERP

Exploration of DIR :: Possible research paths

• Research Pathway 1: Understanding the search results

• Research Pathway 2: Understanding the query

• Research Pathway 3: Understanding the data source

Research Pathway 1 vs Research Pathway 2

Research Pathway 2: Understanding the query

• Blindly routing every query to every data source is unacceptable

• Query understanding • Domain classification of query • Intent recognition of query • Semantic labelling of query

• Route only queries that are relevant to the data source, to the data source: e.g. a News related query to a News source, academic queries to academic sources

• State of the art targets building statistical machine learning methods to solve the query understanding problem

• Include results from data source with SERP

Research Pathway 1: Understanding search results• Blindly routing every query to every data

source is unacceptable

• Understand the search results for clues to unravel nature of query

• Are Advertisements present • Are Images present • Are pdfs types present

• Route only queries that are relevant to the data source, to the data source: e.g. a News related query to a News source, academic queries to academic sources

• State of the art doesn’t focus on search results

• Include results from data source with SERP

Research Pathway 1: Find discriminative features for “non-scholarly materials domain”

Query lengthPermutation of Pages

Result count

Title match

Images present

HTML resource

News present

Google knowledge entity present

Research Pathway 1: Find discriminative features for “scholarly materials domain”

Query lengthPermutation of Pages

Result count

Title subset match

PDF resources

Notable Absences• Google Knowledge

Entity • News • Ads

Notable Presence• Non HTML

resources (PDF)

Research Pathway 1: What next after finding discriminative features?

• Find a dataset (Done) • NASA NTRS query log for scholarly materials domain (400,000+) • AOL 2006 query logs for non-scholarly materials domain (400,000+)

• Train a classify (Not done)• Given a query and a list of search results. Classify the query as

belonging to one of multiple classes e.g. (Scholarly material)

Research Pathway 2: Heuristic for unsupervised domain classificationOriginal algorithm 1:

• Idea: Given a query and a list of search results, the important terms which co-occur across multiple search results are indicative of the domain of the query.

Query 1: Search Engine URIs List

doc2 <a, a, a, b, b.., c>

2: Generate unigram vectors, remove redundant terms

<a, c, x, y, d, d> <a, p, w, s>docn

<a, b, c> <a, c, x, y, d> <a, p, w, s>

<a, a, a, b, c, c, d, p, s, w, x, y>

3: Sort

<a, a, a> <b> <c, c> <d> <p> <s> <w> <x> <y>

4: Find clusters

Domain Set: P

Original algorithm 1 Example: Possible domains for query “Lionel messi”

• (terms), 10 of 11 pages • (barcelona"., barcellona-granada, barcelon,, barcelon,

barcelona), 9 of 11 pages • (best"., best), 9 of 11 pages • (championship, champion, championship,,

champions..., champions:, championships., champions', championships, championship:, champions.", championship-winning, champions, champions".), 9 of 11 pages

• (city, city)), 9 of 11 pages • (club, club's, club's...), 9 of 11 pages • (consented, considerably, consecutively).,

consecutively,, considered, consent, consistent, conscious, consecutively"., consecutive, considers, consider), 9 of 11 pages

• (everybody, every), 9 of 11 pages • (fc, fc.), 9 of 11 pages • (football, football".), 9 of 11 pages • (game"., game".[370], game), 9 of 11 pages

Relevant domains based on human judgement

Original algorithm 2: Heuristic for supervised domain classification

• Given a set of predefined domains D:

<a, a, a> <b> <c, c> <d> <p> <s> <w> <x> <y>

4: Find clusters

Domain set: P

max( similarity (Pi, Di) )

• Similarity • Naive hybrid similarity (Jaccard/Overlap coefficient) • Word net • Explicit Semantic Analysis

Exploration of DIR :: Summary • Problem

• There exists an inconsistency between between SERP and special collections, thus many relevant resources are not included in SERPs or

• Included too late (e.g. last page)

• Goal • Present resources from multiple unindexed sources with Google

SERP which can be done through a browser plugin

• Research Pathways • Understand the search result and train a model to learn when a

query should be forwarded to a special collection • Understand the query, for example the domain, then forward

only relevant queries to their respective special collections • Include results from special collection with SERP

TEMPORAL COHERENCE OF COMPOSITE MEMENTOS IN WEB ARCHIVES

SCOTT G. AINSWORTH OLD DOMINION UNIVERSITY

AUGUST 5, 2015 OLD DOMINION UNIVERSITY

CONTENTS ■ Motivation

(Appearances can be deceiving) ■ Background ■ Temporal Coherence ■ Research ■ What’s next?

8/5/2015 Scott G. Ainsworth • Status for Herbert Van de Sompel Visit

MOTIVATION

APPEARANCES …

4 http://web.archive.org/web/20041209190926/http://www.wunderground.org/cgi-bin/findWeather/getForecast?query=50593 (now 404, but that's a different story…)

… CAN BE DECEIVING

Root Memento-Datetime: 2004-12-09T19:09:26

CLEAR OR CLOUDY?

QUESTIONS ■ How prevalent is temporal incoherence? ■ Can Temporal Coherence be improved using ■ Multiple archives? ■ Additional memento selection heuristics?

■ How can Temporal Coherence be conveyed?

BACKGROUND COMPOSITE MEMENTOS COHERENCE STATES COHERENCE PATTERNS

COMPOSITE MEMENTO

PRESENTATION STRUCTURE

URI-M0

URI-M1 URI-M2 URI-Mi-1...

URI-Mi URI-Mi+1 URI-Mn...

COHERENCE STATES ■ Prima Facie Coherent

Evidence that the memento existed in its archived state when the root was acquired.

■ Prima Facie Violative Evidence … did not exist ...

■ Possibly Coherent Evidence … might have existed ...

■ Probably Violative Evidence … probably did not exist ...

CONSIDER THIS HTML…

AND THESE RESPONSE HEADERS HTTP/1.1 200 OK Server: Tengine/2.0.3 Date: Mon, 27 Apr 2015 22:03:32 GMT Content-Type: image/jpeg Content-Length: 15632 Connection: keep-alive Memento-Datetime: Tue, 07 Feb 2006 00:58:23 GMT Link: <Memento links deleted...> X-Archive-Orig-server: Apache/1.3.26 (Unix) ApacheJServ/1.1.2 PHP/4.3.4 X-Archive-Orig-etag: "4978-3d10-3e4d822e" X-Archive-Orig-content-length: 15632 X-Archive-Orig-accept-ranges: bytes X-Archive-Orig-date: Tue, 07 Feb 2006 00:58:20 GMT X-Archive-Orig-content-type: image/jpeg X-Archive-Orig-last-modified: ↩︎

Fri, 14 Feb 2003 23:56:30 GMT X-Archive-Orig-connection: close <other headers deleted>

PRIMA FACIE COHERENT

Bracket Pattern: Memento-Datetime + Last-Modified

(yes, Last-Modified is sometimes wrong, but many of those cases can be detected)

PRIMA FACIE COHERENT

Equal Pattern: simultaneous capture (with an optionally tunable “bubble of simultaneity”)

PRIMA FACIE VIOLATIVE

POSSIBLY COHERENT

Closest (or only) memento captured before the root

PROBABLY VIOLATIVE

Closest (or only) memento captured after the root but no Last-Modified (possibly indicating a dynamically generated representations)

TEMPORAL COHERENCE EMBEDDED RESOURCES REPRESENTING COHERENCE

TEMPORAL COHERENCE

2005-05-14

01:36:08

+9 days

+18 days +18 days

+7 months

+2.1 years

EMBEDDED RESOURCES Resource Memento-Datetime Delta Resource Memento-

Datetime Delta

h"p://www.cs.odu.edu. 2005205214.01:36:08. spacer.gif. 2005206201.16:23:10. 18.6.d.

mm_menu.js. 2005205223.02:39:12. 9.0.d. jimcheng.gif. 2005206201.16:37:39. 18.6.d.

style.css. 2005205223.02:39:39. 9.0.d. jsmith.gif. 2005206201.16:58:50. 18.6.d.

gfx2logo2odu2crown.gif. 2005205223.02:39:39. 9.0.d. rmenu_1st_featured_alumni.png. 2005206201.21:21:45. 18.8.d.

ddmenu_ddown.js. 2005205223.02:39:43. 9.0.d. hmenu_college_...2new.png. 2005212221.20:14:25. 7.3.mo.

university.js. 2005205223.02:39:56. 9.0.d. rmenu_1st_upcoming_news.png. 2005212221.20:15:14. 7.3.mo.

rmenu_1st_about.png. 2005206201.13:40:25. 18.5.d. rmenu_1st_upcoming_events.png. 2005212221.21:01:12. 7.3.mo.

rmenu_bo"om_229.gif. 2005206201.14:07:29. 18.5.d. lmenu_1st_resources.png. 2005212228.17:47:41. 7.5.mo.

shadow2bl.gif. 2005206201.14:55:53. 18.6.d. bullet_blue_triangle.gif. 2005212228.19:43:48. 7.5.mo.

ecsbdg.jpg. 2005206201.14:56:17. 18.6.d. logo2cs.gif. 2005212228.19:54:29. 7.5.mo.

shadow2br.gif. 2005206201.15:18:18. 18.6.d. rmenu_1st_featured_student.png. 2007206212.02:36:07. 2.1.years.

gfx2btn2go2dblue.gif. 2005206201.15:34:19. 18.6.d. shadow2b.gif. 2007206221.02:35:17. 2.1.years.

shadow2tr.gif. 2005206201.15:55:57. 18.6.d. shadow2r.gif. 404.Not.Found.

header2right1.gif. 2005206201.16:06:16. 18.6.d.

Embedded Resources 26

Mean Delta 125.9 days

Standard Deviation 207.7 days

Minimum Delta 9.0 days

Maximum Delta 2.1 years

REPRESENTING COHERENCE

THE FULL CHART

Mementos by Delta

-3y -1y 0 1y 2y 3y 4y 5y 6y

2013201220112010200920082007200620052004200320022001

Probably Coherent

rURI-M

Probably Violative

Prima Facie Coherent Prima Vacie Violative

2005-03-10

RESEARCH DATA SET SAMPLING STATISTICS

DATA SET ■ 4,000 sample URI-Rs (JCDL’11 data set) ■ Single and Multiple Archives ■ Two Heuristics: ■ Minimum distance (current default

Wayback behavior) ■ choose closest Memento-Datetime

■ Bracket (proposed here) ■ use combination of Memento-Datetime +

Last-Modified (when available)

SAMPLING & RECOMPOSITION ■ For each sample URI-R (rURI-R): ■ Download available TimeMaps ■ Download a single root Memento per

month ■ For each monthly Memento ■ Extract embedded URI-Rs (eURI-Rs) ■ Download TimeMaps for eURI-Rs ■ Download heuristically-best eURI-Ms ■ Repeat recursively

■ Run each heuristic and single-/multi-archive combination

ROOT URI-R STATISTICS

Root URI-Rs archived 2,756 • 68.9% In multiple archives 1,180 • 29.5% Mean archives per URI-R 1.58 Mean mementos per URI-R 124.57

200 OK 82,425 • 93.6% 503 Service Unavailable 4,444 • 5.0% 404 Not found 583 • 0.7% 403 Forbidden 388 • 0.4% Others 214 • 0.3%

URI-M Status

Archival Data

EMBEDDED URI-R STATISTICS

Embedded URI-Rs 1,623,127 per root URI-M 19.7 Embedded URI-Ms available 1,332,993 • 93.6% per root URI-M 15.1

Not archived 312,641 • 83.9% 404 Not found 44,852 • 12.0% 403 Forbidden 6,116 • 1.6% 503 Service Unavailable 5,442 • 1.5% Others 3,508 • 0.9%

URI-M Failure Reasons

Archival Data

COMPOSITE MEMENTO (ROOT) COMPLETENESS & COHERENCE

Description MinDist Single

MinDist Multi

Bracket Single

Bracket Multi

Mean Complete 76.1% 80.2% 76.2% 80.3% Mean Missing 23.9% 19.8% 23.8% 19.7%

Completeness (and Missing)

MinDist Multi

Bracket Single

Bracket Multi

Mean Prima Facie Coherent 41.0% 40.9% 54.7% 54.6% Mean Possibly Coherent 27.3% 28.7% 12.8% 14.2% Mean Probably Violative 2.5% 5.3% 2.5% 5.3% Mean Prima Facie Violative 5.3% 5.3% 6.2% 6.2%

Coherence

At least 5% of pages can be shown to have temporal violations!

Multiple archives: +completeness, -coherence?

EMBEDDED MEMENTO COHERENCE

MinDist Multi

Bracket Single

Bracket Multi

Prima Facie Coherent 622,565 621,447 864,736 859,625 Possibly Coherent 497,405 466,046 244,104 215,585 Probably Violative 104,376 53,734 104,339 53,694 Prima Facie Violative 100,760 103,662 114,062 117,469

Totals 1,325,106 1,244,889 1,327,241 1,246,373

MinDist Multi

Bracket Single

Bracket Multi

Prima Facie Coherent 47.0% 49.9% 65.2% 69.0% Possibly Coherent 37.5% 37.4% 18.4% 17.3% Probably Violative 7.9% 4.3% 7.9% 4.3% Prima Facie Violative 7.6% 8.3% 8.6% 9.4%

At least 7% of embedded resources are used violatively!

WHAT’S NEXT? EQUALITY & SIMILARITY MINOR & MAJOR VIOLATIONS POLICIES & HEURISTICS CONVEYING COHERENCE

EQUALITY & SIMILARITY

Equality and similarity allow prima facie coherence without Last-Modified

Early results: equality yields < 2% improvement

MINOR OR MAJOR VIOLATIONS? ■ This is a temporal violation. But is it

meaningful?

■ How to judge? ■ Most archives transform HTML ■ Few support export of original file

■ How to measure similarity on binary files?

POLICY & HEURISTIC TRADEOFFS ■ Speed: minimize distance ■ Completeness: query all archives

(not just top k) ■ Accuracy: maximize coherence

CONVEYING COHERENCE

How to scale to > 100 embedded mementos?

How to convey coherence & contributing archive?

WHAT’S NEXT SUMMARY ■ Equality & Similarity ■ Significance of violation (major? minor?) ■ Policies & Heuristics ■ Conveying Coherence

Progress Report Lulwah Alkwai

Presented to: Dr. Herbert Van de Sompel

Previous Work

JCDL 2015 Paper: “How Well Are Arabic Websites Archived?” Lulwah M. Alkwai, Michael L. Nelson, and Michele C. Weigle

We won “Best Student Paper Award”

English sports websites are more archived than Arabic

www.espn.go.com www.kooora.com

GeoIP only ccTLD only

Both Neither

!  News: alarabiya.net !  ccTLD: Not Arabic (.net) !  GeoIP: Not Arabic country (US)

!  E-Marketing: haraj.com.sa !  ccTLD: Arabic (.sa) !  GeoIP: Not an Arabic country (Ireland)

!  News: al-watan.com !  ccTLD: Not Arabic (.com) !  GeoIP: Arabic country (Qatar)

!  Educational: uoh.edu.sa !  ccTLD: Arabic (.sa) !  GeoIP: Arabic country (SA)

How do we classify Arabic websites? 4

Selecting seed URIs Name Registered Year URI count

DMOZ US 1999 Dmoz.org/world/arabic 4,086 Raddadi Saudi Arabia 2000 Raddadi.com 3,271 Star28 Lebanon 2004 Star28.com 8,386 Total 15,743

•  15,092 unique seed URIs •  11,014 URIs that existed in the live web

~41% ~38%

~36% ~39%

Language test intersection testing for Arabic language

Total Arabic URIs Dataset = (7,976+292,670) = 300,646

Crawling Arabic seed URIs 7

Findings Our Arabic language dataset was not largely located in Arabic countries

"  Only 14.84% had an Arabic ccTLD "  Only 10.53% had a GeoIP in an Arabic country "  Popular Western domains (e.g., cnn.com, wikipedia.org) appeared in

the top 10 Arabic webpages are not particularly well archived or indexed

"  46% were not archived "  31% were not indexed by Google

An Arabic webpage is more likely to be... "  indexed if it is present in a directory "  archived if it is present in DMOZ "  archived if it has neither Arabic GeoIP nor Arabic ccTLD

For right now, if you want your Arabic language webpage to be archived, host it outside of an Arabic country and get it listed in DMOZ

Youssef Eldakar Bibliotheca Alexandrina

"  Since 2011, the BA crawls have focused on Egyptian content

"  Seeds are manually selected "  Future plans are to cover content related to the Arab

world 9

Bibliotheca Alexandrina

Current Work Replacements for missing images

Goal: Make contribution by finding missing images through context and discover the replacement for the image Example:

Motivation "  D-Lib Magazine, Jan 2005:

“Transparent Format Migration of Preserved Web Content” David S. H. Rosenthal, Thomas Lipkis, Thomas S. Robertson, and Seth Morabito

"  The main idea was to change a file format that is no longer understandable to a new format without changing the URI

"  Can this be done for images with 404 responses? "  We can define a new response code, location header

e.g. “210 Not Quite OK, But Close”

Sample log query 0.36.125.141)web.archive.org)5)[01/Jan/2011:01:30:58)+0000])"GET)hBp://web.archive.org/web/20110101013058/hBp://www.slaverymuseum.org/IraAtTable.jpeg)HTTP/1.1")404)2135)"hBp://web.archive.org/web/20030413174118/www.slaverymuseum.org/home.htm")"Mozilla/5.0)(Windows;)U;)Windows)NT)5.1;)en5US))AppleWebKit/534.10)(KHTML,)like)Gecko))Chrome/8.0.552.224)Safari/534.10")TCP_MISS:SOURCEHASH_PARENT/207.241.227.95)205)

Check full URI in the IA

>"curl"'I"http://web.archive.org/web/20110101013058/http://www.slaverymuseum.org/IraAtTable.jpeg""HTTP/1.1"404"Not"Found"

Server:"Tengine/2.1.0"

Date:"Tue,"04"Aug"2015"18:17:46"GMT"Content'Type:"text/html;charset=utf'8"

Connection:"keep'alive"

set'cookie:"wayback_server=73;"Domain=archive.org;"Path=/;"Expires=Thu,"03'Sep'15"18:17:45"GMT;"

X'Archive'Wayback'Runtime'Error:"ResourceNotInArchiveException:"http://www.slaverymuseum.org/IraAtTable.jpeg"was"not"found"X'Archive'Wayback'Perf:"{"IndexLoad":144,"IndexQueryTotal":144,"RobotsFetchTotal":2,"RobotsRedis":1,"RobotsTotal":2,"Total":390}"

X'Archive'Playback:"0"

URI requested

Referring URI

Check full URI in the live web

">"curl"'I"http://www.slaverymuseum.org/IraAtTable.jpeg"

HTTP/1.1"404"Not"Found"Date:"Tue,"04"Aug"2015"18:15:34"GMT"

Server:"Apache"Content'Type:"text/html;"charset=iso'8859'1"

Check Timetravel

Check domain in the live web

>"curl"'I"http://www.slaverymuseum.org"HTTP/1.1"301"Moved"Permanantly"

Date:"Tue,"04"Aug"2015"18:26:41"GMT"Server:"Apache"

Location:"https://vimeo.com/search?q=slaverymuseum.org"Content'Type:"text/plain;"charset=UTF'8"

Check image name in new page "  Not found

Check leaf page for image name

"  Not found

Check domain in the IA

Check search engine for image surrounding text

"  Using the “src” and saving the “alt” in HTML (alternative information) as a back up.

e.g. "  Image src="IraAtTable.jpeg” "  alt="Ira)Hunter,)Jr.)and)Oni)Lasana

<img)border="0")src="IraAtTable.jpeg")width="120")height="97")align="top")alt="Ira)Hunter,)Jr.)and)Oni)Lasana)">)

Searching Google for (IraAtTable.jpeg)

Found same src name and parts of the surrounding text

http://signhom.net/professionalshub/wp-content/uploads/sites/3/2013/11/IraAtTable.jpg

>"curl"–I"http://web.archive.org/web/20110101013058/http://www.slaverymuseum.org/IraAtTable.jpeg""

210"Not"Quite"OK,"But"Close"

Date:"Wed,"05"Aug"2015"12:56:03"GMT"Location:"http://signhom.net/professionalshub/wp'content/uploads/sites/3/2013/11/IraAtTable.jpg"

New response code

Summary of approaches

"  Check full URI in the live web "  Check full in URI the IA "  Check full in URI the timetravel "  Check domain in the live web "  Check domain in IA "  Check images in the redirected webpage "  Check leaf pages "  Check surrounding text in search engines "  Compare results of different search engine using image

duplication, such as Google large-scale analysis of images: http://googleresearch.blogspot.com/2015/06/inceptionism-going-deeper-into-neural.html

Other ideas Image de-duplication

"  JCDL 2015: “Identifying Duplicate and Contradictory Information in Wikipedia”, by Sarah Weissman, Samet Ayhan, Joshua Bradley, Jimmy Lin

"  Can we do the same for the archives by detecting and removing duplicate images

"  How many duplicate images? "  Which version should be kept?

3URILOLQJ�:HE�$UFKLYHV

6DZRRG�$ODP�DQG�0LFKDHO�/��1HOVRQ&RPSXWHU�6FLHQFH�'HSDUWPHQW��2OG�'RPLQLRQ�8QLYHUVLW\

1RUIRON��9LUJLQLD��

+HUEHUW�9DQ�GH�6RPSHO��/\XGPLOD�/��%DODNLUHYD��DQG�+DULKDU�6KDQNDU/RV�$ODPRV�1DWLRQDO�/DERUDWRU\��/RV�$ODPRV��10

'DYLG�6��+��5RVHQWKDO6WDQIRUG�8QLYHUVLW\�/LEUDULHV��6WDQIRUG��&$

0HPHQWR�$JJUHJDWRU

/RQJ�7DLO�RI�$UFKLYHV

Ɣ ��%��ZHE�SDJHV�DW�,$�GR�QRW�FRYHU�HYHU\WKLQJ

Ɣ 7RS�WKUHH�DUFKLYHV�DIWHU�,$�SURGXFH�IXOO�7LPH0DS��RI�WKH�WLPH��$O6XP�HW�DO��73'/��

Ɣ 7DUJHWHG�FUDZOVƔ 6SHFLDO�IRFXV�DUFKLYHVƔ 5HVWULFWHG�UHVRXUFHVƔ 3ULYDWH�DUFKLYHV

$UFKLYH�3URILOH

Ɣ +LJK�OHYHO�VXPPDU\�RI�DQ�DUFKLYHƔ 3UHGLFWV�SUHVHQFH�RI�PHPHQWRV�RI�D�85,�5�

LQ�DQ�DUFKLYHƔ 3URYLGHV�YDULRXV�VWDWLVWLFV�DERXW�WKH�

KROGLQJVƔ 6PDOO�LQ�VL]HƔ 3XEOLFO\�DYDLODEOHƔ (DV\�WR�XSGDWH�DQG�SDUWLDOO\�SDWFKƔ 8VHIXO�IRU�0HPHQWR�TXHU\�URXWLQJ�DQG�RWKHU�

WKLQJV

$YDLODEOH�3URILOLQJ�5HVRXUFHV

Ɣ &OLHQW�UHTXHVWƔ $UFKLYH�UHVSRQVHƔ $UFKLYH�LQGH[��&';�ILOHV�

$�&OLHQW�5HTXHVW

$Q�$UFKLYH�5HVSRQVH

$�&';�6QLSSHW

3URILOLQJ�6WUDWHJLHV

Ɣ &RPSOHWH�85,�5�3URILOLQJ��85,�5� ��3URILOH�.H\�ż EEF�FR�XN�LPDJHV�ORJR�SQJ"Z ��ż FQQ�FRP��"LG ��

Ɣ 7/'�RQO\�3URILOLQJ��7/'� ��3URILOH�.H\�ż FRP��ż XN��

Ɣ 0LGGOH�*URXQGż XN�FR��ż XN�FR�EEF��LPDJHVż XN�FR�EEF��ż FRP�FQQ��DU

)UHTXHQF\�0HDVXUHPHQWV

&';-�6HULDOL]DWLRQ

85,�.H\�*HQHUDWLRQ

3URILOH�0HUJLQJ

%DVH�SURILOH

1HZ�SURILOH

0HUJHG�SURILOH

'DWDVHW

Ɣ 7KUHH�DUFKLYHVƔ )RXU�VDPSOH�TXHU\�VHWVƔ ��SURILOHV�IRU�HDFK�DUFKLYH�DQG�VDPSOH�VHW

$UFKLYHV

$UFKLYH 85,�5V 85,�0V 6L]H

$UFKLYH�,W ��% ��% ��7%

8.:$ ��% ��% ��7%

6WDQIRUG ��0 ��0 ��*%

6DPSOH�4XHU\�6HWV

6DPSOH ,Q�$UFKLYH�,W ,Q�8.:$ ,Q�6WDQIRUG

'02= ��

0HPHQWR3UR[\ ��

,$:D\EDFN ��

8.:D\EDFN ��

6DPSOH�6L]H��0�85,V�(DFK

(YDOXDWLRQ

Ɣ 5HODWH�&';�6L]H��85,�0��85,�5��DQG�85,�.H\

Ɣ $QDO\]H�SURILOH�JURZWKƔ (VWLPDWH�5HODWLYH�&RVWƔ (YDOXDWH�5RXWLQJ�3UHFLVLRQ�YV��5HODWLYH�&RVW

&';�6L]H�YV�85,�0��8.:$��<HDUV�

$OSKD��E\WHV�SHU�&';�OLQH

85,�0�YV�85,�5��8.:$��<HDUV�

*DPPD�� .��%HWD��

6SDFH�&RVW��8.:$��<HDUV�

3KL��H��

7LPH�&RVW��8.:$��<HDUV�

7DX��H��H��&';��*%85,�0V��085,�5V��07LPH��KRXUV

5HVRXUFH�5HTXLUHPHQW

$UFKLYH�,W

6WDQIRUG

&RVW�YV�3UHFLVLRQ

*URXS &RVW 3UHFLVLRQ

*��+�3��7/'� %RXQG�E\��RI�7/'V ��

*��+�3��''RP��'6XE��'3WK��'4U\� �� §�� *�

*��',QL� §�� *� §�� *�

*��+[3�� §�� *� §�� *�

*��+LJKHU�+P3Q� �� 1RW�([SORUHG

*��85,5� ��

)XWXUH�:RUN

Ɣ *HQHUDWLQJ�VDPSOH�85,�VHWVƔ 3URILOLQJ�YLD�VDPSOLQJƔ /DQJXDJH�SURILOHVƔ (YDOXDWLRQ�RI�FRPELQDWLRQ�SURILOHV�VXFK�DV�

85,�.H\�DORQJ�ZLWK�'DWHWLPHƔ 3URILOHV�IRU�XVDJH�RWKHU�WKDQ�0HPHQWR�

URXWLQJ��VXFK�DV��VLWH�FODVVLILFDWLRQ�EDVHG�SURILOHV��H�J��QHZV��ZLNL��VRFLDO�PHGLD��EORJ�HWF��

&RQFOXVLRQV

Ɣ *HQHUDWHG�SURILOHV�ZLWK�GLIIHUHQW�SROLFLHV�IRU�WZR�DUFKLYHV

Ɣ ([DPLQHG�FRVW�SUHFLVLRQ�WUDGHRIIV�RI�YDULRXV�SROLFLHV

Ɣ 5HODWHG�&';�6L]H��85,�0��85,�5��DQG�85,�.H\

Ɣ *DLQHG�XS�WR��URXWLQJ�SUHFLVLRQ�ZLWK��UHODWLYH�FRVW�ZLWKRXW�DQ\�IDOVH�QHJDWLYHV

Ɣ &RGH�#�*LW+XE��RGXZVGO�DUFKLYHBSURILOHU

What has Justin been up to, lately?

Justin F. BrunellePresentation for Herbert Van de Sompel

08/06/2015

A simpler time...

Mass hysteria. Human sacrifices. Dogs and cats living together.

Missing resources (bad) and Temporal violations (worse)

http://ws-dl.blogspot.com/2012/10/2012-10-10-zombies-in-archives.html

http://en.wikipedia.org/wiki/Main_Page January 18th, 2012

http://web.archive.org/web/20120118110520/http://en.wikipedia.org/wiki/Main_Page:

January 18th, 2012

Not all tools can crawl equally

Live Resource PhantomJS Crawled

Heritrix Crawled, Wayback replayed

CurrentWork4ow• Dereference URI-Rs• Archive • representation• Extract embedded • URI-Rs• Repeat

Proposed Workflow

Current workflow not suitable for deferred representations

Use PhantomJS to run JavaScript, interact with the representation

Two-tiered crawling approach to optimize performance

Current workflow not suitable for deferred representations

Use PhantomJS to run JavaScript, interact with the representation

Two-tiered crawling approach to optimize performance

More URI-Rs in the crawl frontier

Runs more slowly but more deeply

Run-time & Frontier size PhantomJS vs. Heritrix

To appear: iPres2015

Constructed a classi=er for Deferred Representations

Performance metrics of a two-tiered crawling approach

The classi=er helps crawl deferred representations most e>ciently

Current & Future Work

Using PhantomJS to execute actions on the client

– Pushing buttons

– Selecting drop-downs

– Archiving resulting representation changes

Represent representation state in WARCs

– Graph structure of embedded resources

– Replay in the Wayback Machine

http://ws-dl.blogspot.com/2015/06/2015-06-26-phantomjsvisualevent-or.html

Presented(by(Mat(Kelly(for(Herbert(Van(de(Sompel(

Web$Science$and$Digital$Libraries$Research$Lab$Old(Dominion(University,(Norfolk,(VA(

August(6,(2015(

•  Software as a support vehicle

•  Issues investigating for PhD research topic

•  Sample access patterns mitigated by new Memento-related entities

HVDS(PresentaFon( 2(

Building Software as a PhD Researcher

SoGware(as(a(Support(Vehicle(

•  Purpose: capture what user sees into WARC –  instead of delegation-by-URI

•  Barriers: – Restrictive browser extension API (Evolved/time) – Wheel inventing (nothing for WARCs in JS)

•  Perks: – Seeded private web archiving research – Exposed hard-to-archive content

Website:$hKp://warcreate.com(

Blog:$hKp://wsOdl.blogspot.com/2013/07/2013O07O10OwarcreateOandOwailOwarc.html(

•  “Glue” between institutional tools – hard to configure and use

•  Native binaries – difficult to maintain but novel

•  Further facilitated private web archiving interest

Website:$hKp://matkelly.com/wail(

Blog:$hKp://wsOdl.blogspot.com/2013/07/2013O07O10OwarcreateOandOwailOwarc.html(

•  Integrates live + archived web experience

•  Become familiar with Memento dynamics & usage patterns

•  Provide eventual hook into new entities

Website:$hKp://matkelly.com/mink(

Blog:$hKp://wsOdl.blogspot.com/2014/10/2014O10O03OintegraFngOliveOand.html(

•  Given same input (URI), tools produce varying output

•  Experiment to measure variance

•  Identified hard-to-archive resources

•  Highlighted cutting edge browser-crawler �

Website:$hKp://acid.matkelly.com(

Blog:$hKp://wsOdl.blogspot.com/2014/07/2014O07O14OarchivalOacidOtest.html(

Current Research

private(

archive(

private(

archive(

other(

private(

archive(

other(

private(

archive(

private(

archive(

private(

archive(

other(

private(

archive(

TimeMap

other(

private(

archive(

t = k! t = k-1!≠

90 DAYS AT A TIME

ONLY BACK TO ONE YEAR!

1(year(ago( 2(year(ago( 10(year(ago(

180(days(ago(

TimeMap

private(

archive(

Facebook.com$replay$

What(is(expected( What(the(tools(captured(

Internet Archivepublic, aggregated

Archive.todaypublic, aggregated

Foo Archivespublic, non-aggregated

My web archiveprivate, non-aggregated

time →Archives capturingMy homepage

Changes tomy homepage

Internet Archivepublic, aggregated

Archive.todaypublic, aggregated

Foo Archivespublic, non-aggregated

My web archiveprivate, non-aggregated

time →Archives capturingMy homepage

Changes tomy homepage

Sample Access Patterns

OR$TimeMap

•  More mementos from a superset of sources

TimeMap

•  Patterns 1 and 2 are status quo – provided by framework

•  Querying web archives currently only considers public web content – URI for lookup

•  Framework introduces 2 new entities –  Memento Meta Aggregator (MMA)

–  Private Web Archive Adapter (PWAA)

•  Functional superset of (MA)

•  Can act as intermediary client to relay MA results to ultimate user

•  Allows just-in-time (JIT) inclusion of archives – as specified at query time

•  Set of archives aggregated can be dynamic – e.g., Results must not include IA

MY$CAPTURES$

MY$BANK$CAPTURES$

Various(public(web(archives(

My(web(archives(

MY$CAPTURES$

MY$BANK$CAPTURES$

MY$CAPTURES$

MY$BANK$CAPTURES$

MY$CAPTURES$

MY$BANK$CAPTURES$

NOT$AGGREGATED$

Access(via(the(Meta(Aggregator(

MY$CAPTURES$

MY$BANK$CAPTURES$

140(140(

MY$CAPTURES$

MY$BANK$CAPTURES$

Access(via(the(Meta(Aggregator(

…allows(our(archives(to(be(included(

140(155(

HVDS(PresentaFon(

MY$CAPTURES$

MY$BANK$CAPTURES$

140(155(

MY$CAPTURES$

MY$BANK$CAPTURES$

Bob’s$public$CAPTURES$

The$organizaLon’s$public$CAPTURES$1$

The$organizaLon’s$public$CAPTURES$2$

contains$A$B$C$D$

Contains$B$C$D$

Contains$C$D$

•  Allow dynamic and JIT set of archives •  Superset can be recursively constructed •  Sets can be shared

My public captures!can be integrated !

with public web archives’!HVDS(PresentaFon( 34(

•  Regulates access to Private Web Archives (PWAs)

•  Acts as token authorizer

•  With correct credentials, relays results as if querying the PWA directly

MY$CAPTURES$

MY$BANK$CAPTURES$

GET(TOKEN(for(PWA(

Key:(abcd1234(

HVDS(PresentaFon(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

GET(TOKEN(for(PWA(

Key:(abcd1234(

HVDS(PresentaFon(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

ACCESS(OK(

Token:(4f33c64(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

GET(mementos(for(URI(

Token:(4f33c64(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

Token:(4f33c64(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

Token:(4f33c64(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

Token:(4f33c64(OK(

Returning(mementos(

Return(mementos(

For(URI(

3!captures!

10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

TimeMap

HVDS(PresentaFon(

3!captures!

10,000!captures!

10,000

10,000(

10,143(140!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

TimeMapTimeMapTimeMap

HVDS(PresentaFon(

3!captures!

10,000!captures!

10,143(

140!captures!!!3!captures!!!!!10,000!captures!

MY$CAPTURES$

MY$BANK$CAPTURES$

TimeMap

HVDS(PresentaFon(

3!captures!

10,000!captures!

10,143!captures!

... , <http://web.archive.org/web/20150228155703/https://facebook.com/>;rel="memento";

datetime="Sat, 28 Feb 2015 15:57:03 GMT"

, <http://web.archive.org/web/20150228163939/http://www.facebook.com/>;rel="memento";

, <http://web.archive.org/web/20150303162841/https://www.facebook.com/>;rel="memento";

datetime="Tue, 03 Mar 2015 16:28:41 GMT" , <http://users2machine.local/web/20150305000101/https://www.facebook.com/>;rel="memento";

datetime="Thu, 05 Mar 2015 00:01:00 GMT"; key="e395935019ee467c797034ee410cc91e"

, <//wayback.archive-it.org/all/20150305215922/https://facebook.com/>;rel="memento";

datetime="Tue, 05 Mar 2015 21:59:22 GMT"

, <http://previouslyUnaggregated.org/web/20150306123457/https://www.facebook.com/>;rel="memento"; datetime="Wed, 06 Mar 2015 12:34:57 GMT"

datetime="Tue, 10 Mar 2015 14:07:21 GMT" ...

TimeMap

MY$PRIVATE$FACEBOOK$CAPTURES$

NOT RFC 5988 COMPLIANT!

MY$PUBLIC$FACEBOOK$CAPTURES$

MY$CAPTURES$

MY$BANK$CAPTURES$

Token:(4f33c64(

Token:(c5463b4(

GET(TOKEN(for(PWA(

Key:(2265eef3(

No/invalid!token!returned!

Access!denied!or$0!mementos!

HVDS(PresentaFon(

3!captures!

10,000!captures!

MY$BANK$CAPTURES$

Linda’s$Private$Captures$

Bob’s$Private$Captures$

GET(TOKENs(for(PWAs(

Key:(abcd1234,(Archive:(My(

Key:(cab45cbf,(Archive:(Linda$Key:(b0b01b,(Archive:(Bob$

3!captures!

5!captures!

10!captures!

MY$BANK$CAPTURES$

Access(OK(

Token:(7790ca(

Access(OK(

Token:(b0b01b(

ACCESS$DENIED$

3!captures!

5!captures!

10!captures!

MY$BANK$CAPTURES$

Token:(7790ca,((Archive:(My(

Token:(null,(Archive:(Linda$Token:(b0b01b,(Archive:(Bob$

3!captures!

5!captures!

10!captures!

ø(13(

•  Preserve Private Web Content

HVDS(PresentaFon(

•  Simulate & Quickly Deploy Private Web Archives

•  Interface with New Entities Using Memento

New(SoGware:(

•  Background research on state-of-the-art

•  Exploring use cases – Both existing, anticipated, and fabricated

•  Resisting desire to code

HVDS(PresentaFon(

•  Why? – No means exists to integrate private and public

web archives.

•  How to Evaluate? – Does this framework fit real world needs?

Scalable?

•  When will I know I am done? – Any public/private web archive* can be

integrated.

*((((((((((((Ocompliant(

@WebSciDL PhD Student Project Reviews August 5&6, 2015

Internet

Systematic Reviews of Common Chronic pain Medications and Interventions Trang Nguyen MD, PhD MASI 09.24.15 1

@WebSciDL Norfolk, Virginia, USA Web Science and Digital ...mkelly/presentations/2016_tpdl_ipwb.pdf · InterPlanetary Wayback Peer-to-Peer Permanence of Web Archives Mat Kelly, Sawood

Presubmission Proposal Reviews at the College of Nursing (CON) Nancy T. Artinian, PhD, RN, FAAN Associate Dean for Research and Professor

REVIEWS REVIEWS REVIEWS Paleoecology and â€œinter-situ

Optimizing the Format for Presenting Updates of Evidence Reviews: Seeking User Preferences Sydne Newberry, PhD Southern California Evidence-Based Practice

Title_Page_PG · Web viewReviews for Managers: Reviews for Managers: Performance Reviews Overview. Reviews for Managers: Reviews for Managers: Reviews for Managers: Performance Reviews

Acupuncture’s Cardiovascular Actions: A Mechanistic ......REVIEWS Acupuncture’s Cardiovascular Actions: A Mechanistic Perspective John Longhurst, MD, PhD ABSTRACT Over the last

Systematic Reviews and American College of Physicians Clinical Practice Guidelines Amir Qaseem, MD, PhD, MHA, FACP Director, Clinical Policy American College

Title_Page_PG · Web viewReviews for Users: Performance Reviews Overview. Reviews for Users: Reviews for Users: Reviews for Users: Reviews for Users: Reviews for Users: Performance

Epidemiology of chronic non- malignant pain in Spain Professor Jos Kleijnen, MD, PhD Kleijnen Systematic Reviews Ltd Report

REVIEWS REVIEWS REVIEWS A gentle introduction to quantile

Strategic Management PhD Seminar · Strategic Management PhD Seminar MHR 870 ... published paper reviews and one “friendly” review of a classmates ... Printed outline of the proposed

Data Analysis in Systematic Reviews Madhukar Pai, MD, PhD McGill University Montreal Email: madhukar.pai@mcgill.ca

Overview of different types of reviews : Scoping Reviews, Rapid … · 2019. 6. 13. · Overview of different types of reviews : Scoping Reviews, Rapid Reviews, Systematic Reviews

Assistive Technology for Cognition McKay M. Sohlberg, PhD › assets › documents › 20a... · Assistive Technology for Cognition McKay M. Sohlberg, PhD 9 Existing Reviews ATC and

eprints.whiterose.ac.uk · Web viewMultiple risk behaviour interventions: meta-analyses of randomised controlled trials. Nick Meader, PhD, Centre for Reviews and Dissemination, University

Product Recommendation System from Users Reviews using ... · Product Recommendation System from Users Reviews using Sentiment Analysis K. Yogeswara Rao, PhD Assoc Prof, Dept of CSE

Nutrition & Physical Activity Intervention Effectiveness Systematic Reviews & Evidence-Based Recommendations Laurie M. Anderson, PhD, MPH Centers for Disease

REVIEWS REVIEWS REVIEWS Marine seismic surveys and ocean

H2kinfosys Reviews | H2kinfosys Training Reviews