View
230
Download
0
Embed Size (px)
DESCRIPTION
Citation preview
Customizing Statistics for
Sharper Analysis
Laurent Hill
Renaud Garat
April 15, 2013
II-SDV 2013 Nice 2
Companies’
Acceleration
Conceptual Model Of A Patent Based
Analysis
Top
Companies
Technology
Changes
Company
Changes
1.Understandthe OverallLandscape
Strengths:� Despite the Company’s apparently weak patent base, the Company is well
positioned, especially when teamed with Company C
Weaknesses:� Company B is a generalist competitor for the Company to watch
Opportunities:� Company A and Company C are weak generalist competitors that could
represent licensing opportunities for the Company
Threats:� Company A’s patenting efforts are concentrated in the area of Yarn
Constructions, so the Company needs to understand their business plans
SWOT
Analysis
White-Space
Opportunities
University
Opportunities
2.Understand
RecentTrends
4.UnderstandAvailable
Art
5.EstimateFreedom
To Practice
6.Determine WhoTo Watch And
Leverage
Strategic Direction
3.Understand
Close Art
IP Portfolio
Gap Analysis
Geographic
Coverage
Companies’
Focus
Areas
Investment
Rate
Business
Landscape
Companies’
Momentum
Top
Cited-by Art
Similarity
Search
IP Sharks &
Trolls
Data ���� Information ���� Knowledge + Experience ���� Insight
Why Customize Analysis?
• Data quality
• Productivity
• Technical Specific Needs
II-SDV 2013 Nice 3
Why Customize Analysis?
• Data quality
• Names: normalisation – M&A
bank collaterals
• Concepts: normalisation
languages
II-SDV 2013 Nice 4
Why Customize Analysis?
• Productivity
• Guideline with predefined set of charts
• Re-usable templates:
charts, axis, colors, grouping rules (assignees,
concepts, classification codes, inventors…), real-
time interaction (refining/expanding)
II-SDV 2013 Nice 5
II-SDV 2013 Nice 6
Why Customize Analysis?
II-SDV 2013 Nice 7
• Technical Specific Needs
• Concepts grouping / hiding
• Classification codes grouping / hiding
II-SDV 2013 Nice 8
Why Customize Analysis?
• Problems to solve
• Grouping Rules: Permanent vs Contextual
• Technical Use vs Goals: Customized Axis
• Full Text content: Boolean Axis
II-SDV 2013 Nice 9
II-SDV 2013 Nice 10
Why Customize Analysis?
• Conclusion
☺☺☺☺ Powerful and precise
���� Time consuming (system programing?)
? Future…
II-SDV 2013 Nice 11
II-SDV 2013 Nice 12
? Future…
Guided activities
Preconfigured
Analysis Metrics
Case study: building a
technology/use matrix
• The goal is to understand which technologies are used to achieve a given goal
• Corpus : Swimming pool cleaners
• ~ 1100 patents
• Let’s explore a few alternatives on how to do it
II-SDV 2013 Nice 13
First try: concept/concept matrix
• No clear picture emerges, the concepts just reflect important notions in the corpus
II-SDV 2013 Nice 14
Concepts after clean up / Data rules
• Getting better, drilling down on each cell very powerful for understanding
II-SDV 2013 Nice 15
We need a better solution
• Clearly this is not enough, we need to separate problem domain concepts from solution concepts
• Solution: custom axis
• Let’s start by manually sorting out concepts
• One axis for Goals , one axis for Means
II-SDV 2013 Nice 16
Goal / mean matrix, concept
version
• Simple and
clear:
• Good for communication
• Precision and
completeness
could be
improved
II-SDV 2013 Nice 17
Slice & Dice through boolean queries
• We need a sharper tool to dissect the corpus
• We already have a sharp tool with the full text
search engine: fields, proximity…
• Each value in a custom axis can be a search
engine query.
• Example: combining independent claims and
CPC
• ((water 3D (filter??? or purify???))/ICLM or C02F-
2103/42/CPC)
II-SDV 2013 Nice 18
Technology crossing from Custom
analysis axis
19
Another view, using independent claims
II-SDV 2013 Nice 20
Less noise,
some true
white space
emerges
Analyzing white space on
technology crossings
II-SDV 2013 Nice 21
• No
pending
patents
• Most of the
art is
lapsed or
expired
• Dead end?
Case study wrap up
• Editing data, customizing analysis views and axis let us go beyond simple analysis
• By leveraging the precision of a full featured search engine on all patent text, we were able to build a goals/Mean technology matrix
• Domain expertise still needed, no magic bullet
II-SDV 2013 Nice 22
Thank you
Laurent Hill
Renaud Garat
April 15, 2013