Patent data mining and effective patent portfolio...

30 Intellectual Asset Management October/November 2004

An organisation’s patent portfolio forms acritical part of its IP holdings alongside itsdesigns, trademarks, copyrights and tradesecrets. Much of the value from a portfoliocan only be realised through its effectivemanagement. In turn, that requires tools andtechniques to help understand portfoliocontent, how and where this fits in with theorganisation’s competencies and what the

market opportunities are for exploiting thetechnology owned. There is also a need toidentify gaps where complementary technologycan be licensed in and identify non-coretechnology where know-how can be licensedout or divested for financial return. This is theprovince of patent mining.

A clear and effective IP strategy criticallyincorporates a clear and effective strategy formanaging an organisation’s patent portfolio.But what are the key components of aneffective patent portfolio managementstrategy? In broad terms, it comprises fivecomponents: • What do I have? (the IP audit)• What do I need? (the gap analysis)• Acquire what I need (the investment strategy)• Divest what I don’t need (the deployment

strategy)• Ongoing maintenance and monitoring for

effective development of the IP strategyover time

Let’s look at each of these components inturn and highlight some of the research toolsthat help support these stages.

IP audit – what do I have?In order to establish the patent landscape foran organisation, the process begins with anaudit of the intellectual property within theorganisation’s ownership or control. • What patents do I have in my portfolio?• What technologies are they in?• Can I identify my strengths (core

competencies)? • Are there any obvious weaknesses?

This is not simply a question of establishingownership or licensing rights; it includes an

Effective patent mining

Patent data mining andeffective patent portfoliomanagement

There are many tools that help put patent mining theory into practice. Knowinghow to use them can be a key element in developing a commercially successfulpatent management strategy. By Bob Stembridge and Breda Corish

• No. of patents?• Which technologies?• Strengths/weaknesses?• Additional uses?

WHAT DOI HAVE?

WHAT DOI NEED?

ACQUIREWHATI NEED

DIVESTWHATI DON’TNEED

• Who are competitors?• No. of patents?• Which technologies?• Strengths/weaknesses?

• Build on strengths/core competences• Licensing-in opps?• R&D opps?

• Licensing-out opps?• Donation options?• Lapse maintenance?

DEVELOPMENTSTRATEGY

INVESTMENTSTRATEGYGAP ANALYSISIP AUDIT

POWERFUL PATENT SEARCHING AND PATENT ANALYTICS

IP SERVICES

IP STRATEGY DEVELOPMENT

FAST CURRENT AWARENESS + ORIGINAL DOCUMENTS + IP LAW DEVELOPMENTS

Effective patent portfolio management strategy1

assessment of the importance of thetechnology and its likely markets. For keytechnologies, it can also include anassessment of the strength of protectionoffered by specific patents and whether theseare likely to be assertable in the relevantterritories to protect the technology asintended.

As an example of this, when British Telecom(BT) conducted an audit of its portfolio of morethan 15,000 patents, it re-discovered a patentfor hyperlinking technology dating from a timebefore the world wide web was conceived.However, when it tried to assert the rightsembodied in this patent against companiesthat used hyperlinking on the web, a federalcourt ruled that BT was trying to cover toomuch ground, and in late September 2002 BT

abandoned its efforts to uphold the patent. Although BT didn’t win that particular battle,

it still expects to generate annual income ofUS$100million in the next five years fromlicensing its technology and IP rightsenforcement from its wider portfolio3.

Effective research is the key tounderstanding fully what is in a patentportfolio. To illustrate this, let’s examine asimple case study based on the US company,TiVo Inc, providers of streaming digital videorecording products and services. We haveused TiVo as an example because it has acompact portfolio, which makes it easier toillustrate some of the techniques involved.These can, of course, be applied to muchbigger portfolios where the volume of data canpose a real challenge.


Intellectual Asset Management October/November 2004 31

TiVo patent record - includes the patent family2


TiVo’s patent portfolioIn answering the first question in the portfoliomanagement process – what’s in TiVo’s patentportfolio? – we need to consider the distinctionbetween patents and inventions. TiVo hasdeveloped a number of inventions that mayeach be protected by multiple patents.

For our case study, we used Derwent WorldPatents Index® (DWPI®) from ThomsonScientific, as our source. The databasecollates patent documents relating to specificinventions into patent families. A recent DWPIsearch for TiVo Inc as a patent assigneeretrieved 32 records. These correspond to 32unique inventions for which 122 individualpatent publications have been issued inmultiple jurisdictions. Although 29 US patentapplications and granted patents have beenpublished to date, analysis of the priority dataconfirmed that patent protection has beenapplied for in the US for all 32 unique TiVoinventions.

The sample DWPI record reproduced in thisarticle (see page 31) shows a typical inventiondescription and the details of the associatedpatent family. In this instance, patentpublications at the time of the analysis forTiVo’s new software installation methodcomprised a WO/PCT application, a USgranted patent and an Australian application.

Gap analysis – what do I need?Having completed the first step anddetermined what is in a company’s patentportfolio, the next step is to determine whatelse the company needs to get an edge overits competitors. To do this it needs to know: • The identity of its competitors.

• How many patents competitors have in theirportfolios.

• What technology areas these are in.• Where competitors’ strengths lie.• Whether these pose a threat.

Taking TiVo’s portfolio as a starting pointonce more, we can now take a wider look atwho else is patenting in the world of video-and TV-on-demand technology.

TV-on-demand – the global patenting pictureIn order to analyse the landscape for video- andTV-on-demand technology, we first need toisolate the set of inventions relating to thistechnology. A further search of DWPI wascarried out using a combination of keywords andmanual code indexing, specifically for video-on-demand systems and electronic programmeguide systems. This retrieved a set of recordsrepresenting inventions from all players in thismarket space.

A sample of the search results wasexported into a data mining and visualisationtool. Various analyses were generated thatshow key patentees and their patent filingactivity over time.

The resulting illustration shows some of thetop patent assignees in this technology. Theseinclude many familiar names from the TV andvideo industries, illustrating how this approachcan be used to identify both competitors andpossible business partners. This can be ofparticular importance for smaller companiesthat may have new technology but not theresources to bring this to a large customerbase. TiVo, for example, has establishedpartnerships with Sony and Toshiba, among


Top 20 TV on demand patent assignees4


others, to integrate TiVo technology into a hostof new consumer electronics products.

It can also be useful to look further downthe rankings of organisations patenting in aparticular technology sector, in order toidentify the smaller players who may besuitable sources of licensing-in technology orpotential acquisition candidates.

Acquire what I need - the investment strategyHaving gained a better understanding of whereour portfolio sits in the wider market, the nextstep is to try to identify some of the gaps thatneed to be filled to drive growth.• Where can a company increase R&D

investment to build on strengths?• Which key inventors should the company

head hunt to gain competitive advantage?• Where should the company license in

technology to compensate for areas ofweakness?

• Where do competitors’ strengths lie?• Do these pose a threat?• Is there an opportunity to create an alliance?

A patent portfolio doesn’t sit in isolationfrom the rest of the organisation. People are avital part of the technology expertise of anorganisation. “Patents plus people, oftenmakes a more valuable proposition,” saysMike Carr of BT Exact, one of those who havehelped drive the change in IP strategy at BT.

An important piece of information thatpatent data mining can provide is the identityof key individuals involved in developing atechnology. This can be invaluable to anorganisation seeking to enhance its inventivecapacity by recruiting key staff.

At TiVo, the top 10 inventors are: Barton JM(15 inventions); Look H (6); Platt DC (6);Vallone RP (4); Beach B (4); Look HD (4); VanStam W (4); Kao JS (3); Chow CT (3);Waterman A (3).

This analysis is based on data from the 32TiVo inventions identified in the Derwent WorldPatents Index database. The individualinventor names are mapped against thenumber of patent families where those namesappear. Information about the individualsassociated with new inventions of this typecould be of particular value to a companysearching for expertise in the field ofstreaming video technology.

In addition, inventor information can be ofparticular relevance in the case of merger andacquisition activity. For example, in a takeoverscenario, patent analysis can be used toidentify those individuals who are key to thefuture product development pipeline of anorganisation in the post-acquisition phase.

When using patent information in this way, itis always important to put it in the context ofother business information about theorganisation. In this case study, the analysis


Key individuals behind patented technologies5

shows that J Barton is the most commoninventor name on TiVo’s patent applications.Jim Barton is, in fact, one of the originalfounders of the company as well as being itsChief Technology Officer7.

Data mining and visualisation tools can beused to carry out further analysis of theinventions associated with key individuals, andso reveal their particular areas of expertise.

The specific focus of the inventionsassociated with each individual inventor canbe analysed in more detail using theInternational Patent Classifications (IPCs)assigned to each patent publication.

Taking the most common inventor name atTiVo, J Barton, as an example, the IPC listingindicates that his primary area of expertise isin pictorial communications such as television- as represented by IPC H04N. Comparisonswith the IPC profiles of the other TiVoinventors indicates that their patenting activityis focused on different areas, in particularinformation storage and digital informationtransmission as represented by IPCs G11Band H04L respectively.

Divest what I don’t need - the IP strategydevelopmentLicensing out patented technologies is now animportant part of intellectual assetmanagement and one of the key gains to berealised from data mining of patent portfolios.The scale of those gains can be appreciatedby considering the report of a dramatic 10-foldincrease in recent times of global licensingand technology transfer revenues, from US$15

billion in 1990 to US$120 billion in 2002,according to a report by US News and Worldin June 2002.

Specific examples of potential gains includeTexas Instruments which.according to Deloitte& Touche, “now derives more than US$500million of licensing revenues from its unusedpatents”; and IBM, whose licensing revenues,according to the MIT Technology Review from2000, “top US$1.7billion accounting for 15%of the company’s profits”.

Of course, these are seminal examples ofeffective licensing-out strategies, and there arevery few that will be able to come close toequalling such figures. However, they doillustrate a wider truth: there may be potential inany patent portfolio to identify and then exploittechnology for licensing purposes – the rightsowner will never know unless it looks. How thencan we use patent data mining to identifypotential licensees for unwanted technology?

As already discussed, we can use patentanalysis to identify those companies patentingin the same areas of technology and in thisway identify firms that may be interested inlicensing a company’s technology. However, wecan find closer connections between a targetorganisation and other players in the market byusing citations to focus on those patents thatrepresent a common point of interest.

By way of example the table oppositeillustrates the US patent citation relationshipsfor one of the patents from the TiVo portfolio -US6233389 “Multimedia time warpingsystem” – using the Delphion Citation Linkvisualisation tool.



TiVo Inc - patented inventions6

The patent of interest - US6233389 – is inthe middle of the citation map. Those patentcitations identified by the US Patent Officeexaminer for TiVo’s US6233389 patent areshown to the left of US6233389 and thoselater published patents that subsequently citethis invention are shown on the right.US6233389 was subsequently cited in laterpublished US patents filed by Sony, Prediwave,Microsoft, IBM and Micron.

Such a citation analysis has a number ofapplications in addition to the identification oforganisations that may represent potentiallicensees for a technology, such as theidentification of non-obvious competitors andpotential infringement issues.

Ongoing maintenanceDwight D Eisenhower once said: “Plans arenothing; planning is everything.” In otherwords, successful completion of the steps indetermining what is in the starting portfolio,identifying what is missing and what can bedivested, is only part of the process. Ongoingrealisation of rewards from mining a patentportfolio depends on keeping the intelligencebehind the patent portfolio managementstrategy up to date. There are various toolsthat organisations can use for this ongoingmaintenance, including current awarenessservices. These will help to monitor bothcompetitors’ activities and what is happeningin the wider technology area.

ConclusionIn this review, we’ve looked at some of thepatent data mining techniques that cananswer key questions involved in developingan effective patent portfolio managementstrategy. The returns that can be realised bystreamlining areas of competence, as well asdirectly through licensing revenues, aresubstantial, and increasingly contribute asignificant proportion of the earnings oftechnology-based businesses.

The patent data mining techniquesdescribed here, used wisely and in conjunctionwith the effective management of a patentportfolio, can help organisations extract puregold from the nuggets of technology coveredby the patents they own.

Bob Stembridge is Customer RelationsManager at Thomson Scientific in London.Breda Corish is Director at Thomson Scientificin [email protected]@thomson.com



TiVo patent citation map8

Sources:1. Schematic: Thomson Scientific2. Derwent World Patents Index® record on Delphion3. Thomson Scientific “IP Matters” - May 20024,5 & 6. Derwent Analytics™ data mining and

visualisation tool analysis7. TiVo Website www.tivo.com8. Delphion citation tree analysis

Patent data mining and effective patent portfolio...

Documents

Transcript of Patent data mining and effective patent portfolio...