1
ILTA 08 – Search as Strategy
Creating “Information Gravitation” in the Firm
Derek Schueren - Vice President of Business Development, Recommind
Felicity Badcock – Online Services Manager, Mallesons Stephen Jaques
August 2008
Aristotle’s Fantasy
“suppose that every tool we had could perform its task, either at our bidding or itself perceiving the need…that shuttles in a loom could fly to and fro…and a plucker play a lyre of its own accord”
Aristotle
Information Gravitation
Web 2.0Content Linking
The Future of Information Retrieval
Personalization
Visualization
CollaborativeFiltering
2
Search
The Basics of Search: Relevancy
The ability (of an information retrieval system) to retrieve material that satisfies the needs
of the user
Google – Relevancy as a Popularity Contest
3
Google – Page Rank Algorithm
PR(A) = (1-d) + d (PR(T1)/C(T1) + ... + PR(Tn)/C(Tn))
Most popular web pages given highest rank based on 2 inputs:The number of incoming links: The more the betterThe number of outgoing links on the page which points at your page: The less the better
Consumer Search and the Enterprise
Consumer search tools have re-defined what people expectGap between the consumer experience and the enterprise
although the gap is closing in some casesin other cases web search needs to catch up
Attorneys want to “Google it” but expect much more from search in Attorneys want to Google it but expect much more from search in the enterprise than with Google
Questions?
What about the User?Google does not take into account who the user is when determining relevancy of results
What about the content of the query?Google uses Hypertext-Matching Analysis, meaning it uses the text and metadata of the web pages to figure out whether a web page should even appear for your queryweb pages to figure out whether a web page should even appear for your query
How does this relate to search in the Enterprise?
4
Web Search vs Search in Enterprise
The Web contains a huge store of simple content
Relevancy based on key word match over layered with popularity score
Enterprises by comparison have small content stores but much complexity
No link structures mean relevancy needs to be computed in a different way
Web Search Search in Enterprise
y p p y
Internet content typically lacks much context beyond what is linked within its pages
Unrestricted access - ‘public’ content
Non authoritative – near enough is good enough
computed in a different way
Content in the enterprise has many links to information outside of it both inside and outside the enterprise
Enterprise search tools must respect security
How does this relate to search in the Enterprise?
Google – Parallels in the Enterprise
consumer experienceenterprise experience
Google style search boxSearch returned in list formatInformation rich results listSearch terms highlightedAdvanced search & preferences available
Google – Differences in the Enterprise
Information linked together in contextMetadata much richerAbility to filter down post queryRelevancy based on sophisticated content analysis not link structuresRelevancy can extend beyond primary object
consumer experienceenterprise experience
5
Relevancy – Beyond Google
Concept SearchSophisticated content analysis increase relevancy helping people find content even if all the keywords are not present
Content LinkingComplex object types being created through linking of common metadataObjects being fo nd thro gh rele anc joining Objects being found through relevancy joining
PersonalizationMy Search: Based on office, role, position, practice group, industry, etcUser click stream data being used to influence personal relevancy
Collaborative FilteringAggregate click stream data being used to influence general relevancyRecommendations of content based on users with similar interests
Evolution of Enterprise Search
Concept BasedSearch
Key Word Search
Pattern MatchingSearch
Concept Search Example
6
Concept Based Search Systems Automatically Form an Understanding of the Content Being Searched
Concept 3miranda 79.31830
confession 78.94113
tape 74.29385
identification 69.24628
interview 64.72778
interrogation 58.11626
tapes 57.73819
Concept 4patent 99.44380
infringement 98.13741
uspto 90.92023
invention 86.88132
patents 75.83580
copyright 61.92028
software 60.24999
Concept 5environmental 96.64108
water 95.17682
epa 92.34616
waste 88.24619
hazardous 85.70676
pollution 84.56543
disposal 76.36809
photographs 54.45479
pornography 53.98799
conversation 50.49723
statements 49.25360
entrapment 47.14402
told 46.57599
fbi 43.13567
recording 42.54316
statement 41.21378
videotape 39.99082
agent 39.86646
specification 58.69289
equivalents 56.45938
art 54.11242
copyrighted 51.57011
uspq 50.41583
patentee 48.20264
works 47.88340
inventor 44.99235
pto 44.05206
copying 42.91004
patented 42.56361
copyrights 41.80928
infringing 40.52107
cercla 72.48301
clean 70.42545
emissions 52.13936
exxon 50.06532
nuclear 50.05781
cleanup 48.20259
toxic 48.00654
corps 47.04402
contamination 46.52463
asbestos 45.90027
solid 45.40522
sites 45.12626
chemical 44.24835
Content Linking
Importance of content linking in the law firm
Need to find and manage informationefficiency and productivityrisk - regulatory requirement for finding the right information quickly
Traditional systems are being underminedwikis and SharePoint team sites are encroaching on the DMS - increasingly used as repositories for documents storing the ‘working copy’repositories for documents, storing the working copy
Increase in the volume of contentRapidly increasing number of information silos
searching each system is not effective
7
Content Linking Takes Enterprise’s Complexity & Simplifies it for the User
Democratisation of accessback end complexityfront end simplicity
complex back end
simple user interface
Benefits of a Linked Approach
Enterprise search blends content from diverse locations
enterprise search
“internal” collaboration
solutions
punify information silos
as content grows
“external” content
“core” internal systems
Linking Content Provides For a Complete View That is Easy to Get To
8
Results have been filtered down by specific Citation
With Content Linking Both Internal & External Sources
West KM filter allows results to be filtered by citation
All Citations within the document are displayed in the preview screen
For a Much Richer User Experience
Status Flag External Hyperlink
Internal Reference
Personalization
9
Google – Relevancy as a Popularity Contest
Google – What if Relevancy Was All About You?
Joe – Computer Programmer interested in Computer
NetworkingJane – Executive Director interested in
Business Networking
Collaborative Filtering
10
Amazon – Collaborative Filtering
How does Amazon make it easy for me to find what I want?Amazon knows about my past interactions
and the interactions of users who performed similar searches
and ‘hot’ or popular items
Parallels in the Enterprise
Passive presentation of filtered informationDecisiv Email at
Parallels in the Enterprise
System anticipates the user’s needs (Push vs Pull)
Decisiv Email at
filing suggestionscontext sensitive searches
11
Parallels in the Enterprise
Decisiv Email at
filing information
Web 2.0
Content Rating and Social Bookmarking
Yahoo SearchMonkey - ratings for movies, recipes etc – from StumbleUpon
12
Content rating and tagging in the enterpriseMallesons’ Scotty solution – in betapublishes RSS Know How current awareness 'feeds' uses open source technology Drupal as a feed aggregatorincludes tagging and digging
Scotty
items tagged “enterprise 2.0”
Scotty
13
visualization
The visualization of search
Mallesons’ business intelligence systemUser Support call stats
Visualisation – business intelligence
14
Maps - Google ride finder
search for available shuttles, taxis and li i i limousines in your areashows you the actual location of each vehicleonce you have located a vehicle, call the provider to reserve itavailable in 10 US cities
15
Information Gravitation Case StudyInformation Gravitation Case Study
Information gravitation case studyMallesons Find > PeopleFinder
simplein task bar – always thereprefix defines search 1.7 million searches in last 12 months7,000 searches per day
PeopleFinder
16
PeopleFinder pages update in real-time (no manual refresh is required)when a handset is picked up the presence icon changes tousers can control presence information using Office Communicatoricons in use
PeopleFinder presence featuresQuickly scan presence icons to determine availability
PeopleFinder presence featuresHover the mouse on a name to show a rolling 7 day calendar view
a quick glance to check someone’s availabilitycalendar view shows Outlook appointments, flexible working arrangements, training dates/times, leave status and public holidaysno ‘drill down’ into the detail of calendar entriesintegrated time zone support e.g. Sydney / Hong Kongshows if a person is logged into IM on their BlackBerryCommunicator note and Outlook out-of-office messages shownnext available time is highlightedPC idle time is another useful indicator of ‘presence’simple opt-out
PeopleFinder presence featuresRight click on a name for communication options and contact information
send an emailstart an Office Communicator chatdial an extension or mobile phone numberuse camp-onsend a SMS text messageview profile information and office locationdd th t i k t t li tadd the person to your quick contacts list
click on any phone number in PeopleFinder or quick contacts list to make a callsupports extension mobility
17
PeopleFinder presence featuresCamp-on
feature popular in legacy PABX systemsleverages presence information in addition to telephone status e.g. alert triggered by ‘available’ calendar statussimple browser-based popup windowwhen alert is triggered pop-up comes to the foremultiple camp-on requests can be active at oncemultiple camp on requests can be active at once
PeopleFinder presence featuresFloor plans and quick contacts
printable floor plans show office locationdata drawn from ‘moves’ workflow process
PeopleFinder presence featuresPeopleFinder integrated with Cisco Customer Response Application
18
PeopleFinder presence featuresMobile device support
quickly search for staffpresence icon integrationclick to dial any phone numberclick to create new email
PeopleFinder code used to exploit Cisco AXL APIresulted in BlackBerry application to manage call forwarding on VoIP handsets from mobile devices
PeopleFinder presence technologyReal-time Web 2.0 application
Integration ofCisco CallManagerMicrosoft ExchangeMicrosoft Active DirectoryMicrosoft Live Communication Server LCSPeopleSoft (staff data and working hours)Metastorm Workflows (staff leave, public holidays, training, moves, and secondments)Metastorm Workflows (staff leave, public holidays, training, moves, and secondments)SQL Server / Oracle database repositories (eg Cisco CRA, Premises database)Keystone
Information delivered in real-time through web browseruses standards compliant technology – XHTML, AJAX, XML, JSON-Pimplemented as abstract API for use in any web pagedeployed to intranet corporate directory and homepage
Serving 2,200+ users 2 million telephony data feeds generated per day42,000 presence feeds generated per hour
PeopleFinder II - in beta
19
Next steps - extending PeopleFinder to the client
desktop agent
Information Gravitation Trends
Information Gravitation Trends
Breakdown of Information BoundariesEnterprise content & Internet content melding together (LinkedIn, Facebook, Google, Yahoo, Microsoft, etc)
Forging Deeper Client RelationshipsInstant Message, Text, E-mail, Beyond Extranets, etc
Information Gravitation around Information Objects (e.g. Client/Matter)Information Gravitation around Information Objects (e.g. Client/Matter)Client/Matter Gravitation (Docs, People, Orgs, Work Done, News...)
Information AwarenessSystem aware of and anticipates user’s needs (Push vs Pull)
The Importance of the CloudInformation/Applications increasingly moving to the CloudThe Cloud allows for vastly more connections and interactions to occurInformation/Knowledge delivered as a service, anywhere, anytime