Automatic Concept Identification: Extracting Problem Solved Concepts From Patent Documents

1 / 24
About This Presentation
Title:

Automatic Concept Identification: Extracting Problem Solved Concepts From Patent Documents

Description:

Rule-based with tags (regular expressions and proprietary code) Domain area lists ... Search knowledge base using semantic components ... –

Number of Views:63
Avg rating:3.0/5.0
Slides: 25
Provided by: danph6
Category:

less

Transcript and Presenter's Notes

Title: Automatic Concept Identification: Extracting Problem Solved Concepts From Patent Documents


1
Automatic Concept Identification Extracting
Problem Solved Concepts From Patent Documents
  • IRFS 2007 Vienna
  • Information Retrieval Facility Symposium
  • Dr. Daniel J. Phelps,
  • DJP Information Consulting Services, LLC

(c) 2007 - DJP Information Consulting Services,
LLC
1
2
Outline
  • Basics of patents
  • Problem solved concept
  • High value applications enabled
  • What makes concept identification hard
  • Organizations working in the area

2
(c) 2007 - DJP Information Consulting Services,
LLC
3
A Patent Is
  • A documented solution to a technical problem
    using a specified technology
  • Legal document giving an assignee the right to
    stop others from using his/her patent solution
    for a stated period of time
  • Repository of technical know-how

3
(c) 2007 - DJP Information Consulting Services,
LLC
4
Problem Solved Concept
  • Statement of what problem is solved in the
    patent
  • Want to understand what problem a patent covers
  • Makes it easier to understand what the patent can
    be used for
  • Helps determine the value of a patent

There is only supposed to be one problem solved
in one patent, but there are many exceptions to
this rule.
4
(c) 2007 - DJP Information Consulting Services,
LLC
5
What Is Enabled By Availability Of Problem
Solved Concept Information
  • Problem solved RD process for creating products
  • Patent mapping for strategic planning

5
(c) 2007 - DJP Information Consulting Services,
LLC
6
Typical RD Process
Start Project
Identify Problem
Try To Solve Problem
Use Individual Company Knowledge
Eventually Solve Problem
Include In Final Product Design
Check On Status Of IP
IP Clearance Decision?
6
(c) 2007 - DJP Information Consulting Services,
LLC
7
Problem Solved RD Process I
Useable Solution?
Yes
Start Project
Identify Problem
Search Problem Solved Database
A
No
Develop Solution
Include In Final Product Design
Seek IP Coverage
IP Clearance Decision
7
(c) 2007 - DJP Information Consulting Services,
LLC
8
Problem Solved RD Process II
Work Around
Work Around Or License?
Develop Solution
Seek IP Coverage
Include In Final Product Design
License
IP Clearance Decision
Negotiate License
Include In Final Product Design
8
(c) 2007 - DJP Information Consulting Services,
LLC
9
Patent Mapping
  • Problem Solved Technology Applied
  • Strategic analysis of IP

10
What Makes It Hard To Identify Problem Solved
Concept?
  • Multiple ways of expressing the concept of
    problem (problem, objective, etc.)
  • Multiple levels of problems
  • Main problem
  • Sub problem
  • Location in patent varies with patent granting
    authority and document type

10
(c) 2007 - DJP Information Consulting Services,
LLC
11
Structural Clues in US Patent
Summary of the Invention
Title
Abstract
Background of the Invention
Field of the Invention
11
(c) 2007 - DJP Information Consulting Services,
LLC
12
Derwent World Patents Index
  • Online database from Thomson/Dialog
  • Patents/Applications categorized and indexed
    manually
  • Descriptive title
  • Use/Advantage Statements
  • Field searchable

12
(c) 2007 - DJP Information Consulting Services,
LLC
13
Derwent Use/Advantage Statements
Descriptive Title
Use Statement
Advantage Statement
Courtesy of Ron Kaminecki, Dialog Corporation
13
(c) 2007 - DJP Information Consulting Services,
LLC
14
ClearForest
  • Text mining technology
  • Tagging engine
  • Extraction modules
  • Analytics
  • Has patent module
  • Extracts problem solved terms phrases
  • Approach
  • Rule-based with tags (regular expressions and
    proprietary code)
  • Domain area lists
  • Document structure

14
(c) 2007 - DJP Information Consulting Services,
LLC
15
ClearForest
Courtesy of Michael, ClearForest Corporation
15
(c) 2007 - DJP Information Consulting Services,
LLC
16
PatentCafe
  • Online patent search tool
  • Primary search tool
  • Latent Semantic Analysis
  • Secondary search tool
  • Heuristic Advanced Boolean
  • Find related problems when using concept query
  • Do concept search across 12 million patents

16
(c) 2007 - DJP Information Consulting Services,
LLC
17
Sample PatentCafe Concept Search
  • Concept Search Search using the claims of a
    Company A ink jet patent (nozzle designed to
    constrain dispersal pattern of ink for high
    resolution printing)
  • Return a large body of art in fire extinguisher
    patents (nozzle constraining extinguishant for
    managing fires from a distance). In both cases,
    the nozzle geometry and material viscosity were
    similar.

Courtesy of Andy Gibb, PatentCafe
17
(c) 2007 - DJP Information Consulting Services,
LLC
18
Invention Machine Goldfire InnovatorTM
  • Speeds process of innovation
  • Uses natural language processing and
    computational linguistics
  • Breaks each sentence down into semantic
    components and stores them in a searchable
    knowledgebase
  • Search knowledge base using semantic components
  • Most advanced in extracting problem solved
    concepts

18
(c) 2007 - DJP Information Consulting Services,
LLC
19
Invention Machine Goldfire Semantic
Relationships
The actuator raises the door, it comprises a
diaphragm and hydraulic fluid, but a fluid leak
can result in abrasion and failure of the
actuator.
raises
Courtesy of Jeff Boehm, Invention Machine Corpor
ation
actuator
door
Subject
Action
Object
comprises
diaphragm, hydraulic fluid
actuator
Whole
Consists of
Parts
results in
actuator abrasion, actuator failure
fluid leak
Cause
Produces
Effects
19
(c) 2007 - DJP Information Consulting Services,
LLC
20
Accelovation
  • Market Discovery Tool
  • Technology started at MIT Media Lab
  • Crawls deep web and other content looking for
    mention of needs or solutions1
  • 1.7 million solution statements
  • 1.0 million needs statements
  • Uses text analytics to
  • Match needs of exiting markets with new technical
    solutions
  • Match existing technology solutions (benefits)
    with the needs of new markets
  • Not problem solved approach but analogous in
    concept

20
(c) 2007 - DJP Information Consulting Services,
LLC
21
Accelovation Concept
  • Organization A has a need for a way to solve a
    particular problem in their market
  • Organization B has a technical solution that
    would benefit Organization A
  • Company A initiates a search against the
    repository of needs and solutions statements
    found in Accelovations web search
  • The software makes the match between Organization
    As need and Organization Bs solution (benefit)

21
(c) 2007 - DJP Information Consulting Services,
LLC
22
Summary Conclusions
  • A patent database searchable by problem solved
    would enable a much more efficient RD process
  • The task is not easy, but there are several
    organizations working in the area using different
    approaches
  • Invention Machine comes closest to providing this
    service in the patent realm
  • Accelovation sells an analogues service in the
    market discovery realm that could be used with
    patents

22
(c) 2007 - DJP Information Consulting Services,
LLC
23
References
1Seth Grimes, Intelligent Enterprise, May 17,
2007 http//www.intelligententerprise.com/blog/arc
hives/2007
24
Contact Information
  • Dr. Daniel J. Phelps
  • DJP Information Consulting Services, LLC.
  • DJP.Info.CS_at_frontiernet.net

24
(c) 2007 - DJP Information Consulting Services,
LLC
Write a Comment
User Comments (0)
About PowerShow.com