Download Architeture for XML IR in P2P

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Analyzing Document Retrievability in
Patent Retrieval Settings
Shariq Bashir, and Andreas Rauber
DEXA 2009, Linz, Austria, 31 August – 4 September
Department of Software Technology and Interactive Systems
Vienna University of Technology, Austria
{bashir, rauber}@ifs.tuwien.ac.at
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Motivation
Patent retrieval is a emerging & challenging area.
Patents fall into legal category, use to protect inventions.
 Patents are Complex
– Patents have large document length.
– Contain complex vocabulary.
– Contain complex structure and technical contents.
– Patent writers often intentionally use vague words and expressions, in order
to pass their patents from examination test.
– This creates serious word mismatch problems.
– Relevant patents could not be findable from their relevant queries.
– Users (Attorneys, Patent examiners) mostly use hundreds of queries for
 Patent Retrieval is different to Web Retrieval
– Patent retrieval is recall oriented domain.
– Finding all relevant patents is considered more important than finding only small
set of top relevant patents.
• Exp: A single prior-art patent can invalidate the application of new patent,
• but can we find such patent in given retrieval model?
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Motivation
 Role of Retrieval System in Accessing
Information
– Generally, there is always argue on the quality of user queries.
– Therefore, rather than arguing on the quality of user queries.
– In this paper, we check the role of retrieval systems in accessing
information.
–
–
–
–
Can we access all information using given Retrieval Model?
How much retrieval system’s bias restrict our access to information?
Are there some subsets in given collection, which could not be find?
How easily we can find information in given retrieval system?
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Document Retrievability (aka Findability)
 We measure retrieval systems effectiveness using findability
measure.
 Findability Measure
–
–
–
–
–
Measures how easily a retrieval model can find all documents.
Findability is measured with top c results. (e.g. c = 35, c = 80 etc).
Can figure out which retrieval systems is better for finding patents.
Can figure out high/low findable subsets in the collection.
Can figure out non-findable subsets in the collection.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Computing Findability Measure
 Given a collection of documents D with large set of Queries Q.
 The findability of document d1 is, how many times we can
access d in top-c results, with all queries in Q.
 Exp: If a document d1 in findable in top-c of query q1,
findability score r(d1) = 1.
 kdq is the rank of dD in query qQ.
 f(kdq,c) returns a value of 1 if kdq<= c, and 0 otherwise.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Our Contribution
 Findability is measured with
single score across all queries.
 We consider relevance of queries,
analyzing
– Findability across all queries
– Findability considering only queries
that the document is relevant for
– Findability for queries that a
document is NOT relevant for
– Characteristics of high/low findable
documents
– To what extend we can increase the
findability of documents
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Experiment Setup
 Retrieval models used
– TFIDF, BM25, BM25F, Exact Match
 Patents from US Patent and Trademark website
http://www.uspto.gov
 USPC class 433 - Dentistry Domain
 For query generation, we used only Claim section
 For indexing and searching we used all sections
– Title, Abstract, Claim, Background Summary, Description, Captions
 We used cut-off rank factor c = 35.
Total
Patents
Unique
Terms
Average Patent Length
(words)
Average Claim Section
Length (words)
7,213
53,456
2,888
878.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Query Generation
 Queries based on patent invalidity search scenario
 Extract all single terms from individual patents
term frequency > 2 in claim section
 Single terms expanded into two & three term combinations
 A query is considered relevant for patent, if all its terms
appear at least 3 times in a document
Approach
Total Queries
Average Retrievability Score
Single Term Queries
9,751
345.3
2-Terms Queries
67,735
317.6
3-Terms Queries
337,200
248
All Queries
414,686
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Patent Number: 5,348,473
Patent Title: Medical Tool
Queries:
"GEAR", "CYLINDR", "PORTION", "BODI", "DRIVEN", "PROTRUS", "TOOL", "MEDIC",
"RESPECT"
Term Frequency > 2
Claims:What is claimed is:
1. A medical tool, comprising: a housing including an elongated cylindrical body and a
substantially cylindrical head portion transverse to an axis of said cylindrical body;
a drive gear; and a driven gear; wherein at least one of said drive gear and said driven gear
have a diameter greater than an inside diameter of one of said cylindrical body and said head
portion, respectively, said drive gear being disposed in said cylindrical body and said driven
gear being disposed in said substantially cylindrical head portion, and further comprising
protrusions extending from said elongated cylindrical body and said cylindrical head portion,
said protrusions accommodating said drive gear and said driven gear, respectively.
2. A medical tool as recited in claim 1, wherein said drive gear and said driven gear have
beveled faces, and said protrusions conform to the shape of these beveled faces.
3. A medical tool as recited in claim 1, wherein at least one of said cylindrical body and said
head portion have a protrusion surrounding portions of at least one of said drive
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results with Single-Term Queries rank-cut off = 35
BM25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results with Two-Terms Queries rank-cut off = 35
BM25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results with Three-Terms Queries rank-cut off = 35
BM25
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Some Low Findable Patents with BM25
Patent Title
Retrievability
in all Queries
Retrievabilit Total
Retrievability
y in Relevant Relevant in Irrelevant
Queries
Queries Queries
Dental implant for the
securement of fixed
dental prostheses
2
0
37
2
Electric toothbrush with
vibration
6
0
37
6
Dental lining composition
6
0
27
6
Dental implant member
7
0
81
7
Optionally cross linkable
coatings for orthodontic
devices
7
0
81
7
Dental floss
8
0
81
8
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Some High Findable Patents with BM25
Patent Title
Retrievability
in all Queries
Retrievabilit Total
Retrievability
y in Relevant Relevant in Irrelevant
Queries
Queries Queries
Implants, device and
method for joining tissue
parts
2,146
12
81
2,134
Method and system for
2,028
comprehensive evaluation
of orthodontic treatment
using unified workstation
12
81
2,016
Teeth cleaning implement
with integrated fluid
dispenser
1,905
16
81
1,889
Method and apparatus for 1,838
the three-dimensional
registration and display
of prepared teeth
11
81
1,837
Dental filling material
18
81
1,797
1,805
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion
 We analyze patents retrieval with findability measure.
 We differentiate findability using relevant & irrelevant queries.
 Our results indicate that
– With well-known retrieval models, we could not able to find some patents
in top-c results.
– Large retrieval patents are more findable from irrelevant queries than
relevant queries.
– There is lot of noise on Top-c results of queries.
 Future Work
– For handling word mismatch, we need efficient Query Expansion technique.
– Individual patents have different findability scores in different retrieval
models.
– Exp: Patents which are low findable in Model A, are high findable in Model
B.
– We need efficient Fusion technique.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Thank You
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .