fantoni urgo - cirp dictionary

20
The value of structured semantic information in Engineering and Technology Authors: Fantoni, G. (2), Urgo, M. (RA), Dell'Orletta, F., Pavanello T. Presenters: Dr. Gualtiero Fantoni (University of Pisa) Dr. Marcello Urgo (Politecnico di Milano)

Upload: gualtiero-fantoni

Post on 24-Jan-2015

155 views

Category:

Software


3 download

DESCRIPTION

Analysis of technical papers and terms from technical dictionaries (e.g. CIRP dictionary, CIRPedia, etc..). Solution of disambiguation of technical texts through high quality dictionaries.

TRANSCRIPT

Page 1: Fantoni Urgo - Cirp Dictionary

The value of structured semantic information in Engineering and Technology

Authors: Fantoni, G. (2), Urgo, M. (RA), Dell'Orletta, F., Pavanello T.

Presenters: Dr. Gualtiero Fantoni (University of Pisa) Dr. Marcello Urgo (Politecnico di Milano)

Page 2: Fantoni Urgo - Cirp Dictionary

2

Motivation and Thanks

The inspiration for the experiments shown in this presentation came from the Terminology meeting at the January meeting 2014 and the revision phase of the Italian translation of the CIRP Dictionary.

We take the opportunity to thank:• CIRP Terminology Committee: for their curiosity in investigating this

analysis and for their invitation to present the results. • Springer: for providing the CIRP Dictionary in electronic format;• Prof. Tolio for his interest and support in this analysis.• Italian translation team: all the persons involved in the Italian

translation of the CIRP Dictionary.• Prof. Sami Chatti: for providing the revised version of the Vol. 1

together with the new structure, although in progress;• Felice dell’Orletta: CNR, NLP specialist;• Tommaso Pavanello: IT Patent analysts at Erre Quadro srl, spin off

company of the University of Pisa;• Palazzesi Christian and Zigoni Alberto: Elsevier.

Page 3: Fantoni Urgo - Cirp Dictionary

3

A methaphor

The higher the quality of the maps, the lower the possibility of getting lost.

The higher the quality of the course, the higher the possibility of understanding and learning.

Page 4: Fantoni Urgo - Cirp Dictionary

4

Text-to-Knowledge and Knowledge-to-Text

Text-to-Knowledge combines a battery of tools for Natural Language Processing (NLP), statistical text analysis and machine language learning which are dynamically integrated to provide an accurate representation of the domain-specific context of text corpora in different domains. It allow to extract relevant concepts and the map of their relationship.

Knowledge-to-Text is a technique to enrich a text with high value information derived from certified sources. Automatic disambiguation, references, cross linking are key elements in searching and understanding documents and allows users to increase the precision of their queries.

The more structured and complete is the source of information, the higher is the precision of the query. Google, EPO, etc.. are interested in the theme.

Page 5: Fantoni Urgo - Cirp Dictionary

5

Natural Language Process technology

text

Tokenizer

Morphological Analyzer

PoS Tagger

Chunk extractor

Dependency Parser

annotated text

Sentence Splitter The Natural Language Process technology is used to extract:

Named Entities

Semantic relations

Domain-relevant entities

Page 6: Fantoni Urgo - Cirp Dictionary

6

From text to knowledge and … viceversa

Dealing with technical texts, we need to take into account: • the extraction of terms corresponding to domain-relevant concepts • the identification of the specific domain they refer to (i.e. papers on a

particular subject or the legal domain e.g. patents)

We adopted • a contrastive approach to entity extraction

The domain relevance of entities is assessed on the basis of the contrastive distribution of relevant candidate entities across an input corpus and a different corpus

The contrastive analysis is iterated twice:

1. against a top list of open-domain entities (e.g. from newspapers) to prune common entities (e.g. following day)

2. against a top list of entities from e.g. a different regulated domain (e.g. papers belonging to different subjects)

Page 7: Fantoni Urgo - Cirp Dictionary

7

Experiments

We applied the NLP technology on CIRP knowledge base and in particular we performed 3 experiments:

1. CIRP dictionaries to discover missing definitions (if any).

2. CIRP papers from STC-F to find missing definitions (single and multiwords).

3. Scopus indexed and author keywords1 to generate a Researcher’s profile.

1 the possibility of accessing many of the data in Scopus have been provided by Elsevier through the API service. In particular the authors would like to thank Palazzesi Christian and Zigoni Alberto.

Page 8: Fantoni Urgo - Cirp Dictionary

8

CIRP DictionaryPotentially missing terms (1)

CIRP Dictionary

Definitions in English

Not in CIRP Dictionary

and not generic terms

Missing terms (?)

Page 9: Fantoni Urgo - Cirp Dictionary

9

CIRP DictionaryPotentially missing terms (2)

The CIRP Dictionary is tailored to a specific area, hence, it may not contain really general terms that can be considered as generally agreed in a given language:

surface, process, machine, etc.

Nevertheless it could be worth taking into consideration some terms:• Electrode: frequently used in lemmas and definitions in Vol. 2 and 4;• Grinding wheel: frequently used in lemmas ( e.g., grinding wheel

radial wear) and definitions in Vol. 2;• Feed direction: frequently used in definitions in Vol. 2;• Cutting tool: frequently used in lemmas ( e.g., combination cutting

tool) and definitions in Vol. 2;• Insert: frequently used in lemmas ( e.g., brazed insert) and definitions

in Vol. 1,2 and 4;• Working roll: used in definitions in Vol. 1;• …

Page 10: Fantoni Urgo - Cirp Dictionary

10

CIRP DictionaryEnriching the knowledge (1)

CIRP STC-F

Definitions in EnglishNot in CIRP Dictionary

New terms (?)

210 Papers

2002-2014

Page 11: Fantoni Urgo - Cirp Dictionary

11

CIRP DictionaryPotentially missing terms (2)

The candidate terms could identify new trends of the research in the Metal Forming area, or specific terms whose use has become so common to take into consideration their presence in the Dictionary...

supporting rolllubricant escapepunched sheetforming stepplane strainboss heightaccumulative roll bondingstrain distributionsplitting rollpress hardeningpunching loadtextured rolldummy sheetconical punch

plastic anisotropypulsating hydroformingsubsequent yield locussurface graindeep drawing testfracture zoneindentation pressurebending testfriction modelskin pass rolling….

.. or put reference to other dictionaries or encyclopedias.

.

Page 12: Fantoni Urgo - Cirp Dictionary

12

Search using semantic structured informationExample – People’s research fingerprint

Dr. Matteo STRANO # Bending (+Tube bending+draw bending) 15 Rotary draw bending 6 Tube hydroforming 6 Roll forming 3 Sheet forming processes 4 metal foam 13 FEM 13 Optimization 11 Numerical approaches 6 Objective functions 3 Process condition 11 Reliability 6 Metamodeling 3 Uncertainty 3

Dr. Marcello URGO # Production Planning 18 Production process 10 Scheduling 7 Stochastic methods 7 Virtual factory 6 Manufacturing-to-order 5 Project scheduling 5 Material requirement planning 4 Precedence relations 3 Production system 3 Reconfiguration 3 Material procurement 2 Performance Evaluation 2

More than 50 keywords belong to the CIRP Dictionary Vol.1 on Metal Forming

0 keywords belong to the CIRP Dictionary Vol.1 on Metal Forming, no match with Matteo Strano

Page 13: Fantoni Urgo - Cirp Dictionary

13

Search using semantic structured informationExample

We look for documents and resource related to the production of tubes in metal through a forming process.

Looking into the CIRP Dictionary Vol. I, we find two occurrences of the term “tube”, one in the “Rolling” subchapter and one in the “Drawing” subchapter.

tube: product open at both ends with a circular or polygonal cross-section.

1: Metal Forming5: Rolling

6: Rolled components and properties

tube: product with a round or a polygonal cross section open at both ends.

1: Metal Forming6: Drawing

6: Drawn components and properties

Page 14: Fantoni Urgo - Cirp Dictionary

14

Search using semantic structured informationExample – Documents (1)

Source: Scopus

Query: TITLE-ABS-KEY(tube) AND (LIMIT-TO(DOCTYPE, "ar"))

Order By: Relevance Date: 18 August 2014

1. Tube side performance of new efficient composite enhanced heat exchanger, Zhu, D., An, D., Li, X.,, Zhu, H., Yu, T., 2014, Huagong Xuebao/CIESC Journal, 65 (2), pp. 453-459.

2. Tube current reduction in pediatric non-ECG-gated heart CT by combined tube current modulation, Goo, H.W., Suh, D.S., 2006, Pediatric Radiology.

3. Tube introducer catheter as an adjunct to the airway scope for tracheal intubation in a manikin model, Iizuka, T., Shimoyama, N., Notoya, A., 2010, Japanese Journal of Anesthesiology

4. Tube support effectiveness and wear damage assessment in the U-bend region of nuclear steam generators, Boucher, K.M., Taylor, C.E., 1996, American Society of Mechanical Engineers, Pressure Vessels and Piping Division (Publication)

5. Tube worms promote community change, Callaway, R., 2006, Marine Ecology Progress Series

Page 15: Fantoni Urgo - Cirp Dictionary

15

Search using semantic structured informationExample – Documents (2)

Source: Scopus

Query: (TITLE-ABS-KEY(tube) AND ALL(rolling) AND ALL(metal forming)) AND (LIMIT-TO(LANGUAGE, "English")) AND (LIMIT-TO(DOCTYPE, "ar"))

Order By: Relevance Date: 18 August 2014

1. Control of change in transverse wall thickness variation during batch rolling of tubes, Popov, M.V., Khaustov, G.I., Vol'fovich, G.V., Furmanov, V.B., Rasin, G.V.,1989, Steel in the USSR, 19 (1), pp. 33-36.

2. Three dimensional thermo-mechanical simulation of the tube forming process in Diescher's mill, Pater, Z., Kazanecki, J., Bartnicki, J., 2006, Journal of Materials Processing Technology.

3. Explorative study of tandem skew rolling process for producing seamless steel tubes, Wang, F.-J., Shuang, Y.-H., Hu, J.-H., Wang, Q.-H., Sun, J.-C., 2014, Journal of Materials Processing Technology.

4. Optimization of Cold Rolling of Precision Tubes, Huml, P., Fogelholm, R., Salwén, A., 1993, CIRP Annals - Manufacturing Technology.

5. Present and future developments of metal forming: selected examples, Voelkner, W., 2000, Journal of Materials Processing Technology.

Page 16: Fantoni Urgo - Cirp Dictionary

16

Search using semantic structured informationExample – Documents (3)

Source: Scopus

Query: (TITLE-ABS-KEY(tube) AND ALL(drawing) AND ALL(metal forming)) AND (LIMIT-TO(LANGUAGE, "English")) AND (LIMIT-TO(DOCTYPE, "ar"))

Order By: Relevance Date: 18 August 2014

1. A conical mandrel tube drawing test designed to assess failure criteria, Linardon, C., Favier, D., Chagnon, G., Gruez, B., 2014, Journal of Materials Processing Technology, 214 (2), pp. 347-357.

2. Numerical and experimental analysis of tube drawing with fixed plug, Neves, F.O., Button, S.T., Caminaga, C., Gentile, F.C., 2005, Journal of the Brazilian Society of Mechanical Sciences and Engineering.

3. Investigation of zipper defects in the floating mandrel drawing of small diameter copper tubes, Damodaran, D., Wibowo, F., Shivpuri, R., 1996, Technical Paper - Society of Manufacturing Engineers.

4. Effects of anisotropy in drawing and extrusion processes of bulk metal forming, Pöhlandt, K., Lange, K., Zucko, M., 2006, Steel Research International.

5. Description of a mathematical model of deformability for the process of drawing tubes on a fixed mandrel, Pospiech, J.,1998, Journal of Materials Engineering and Performance.

Page 17: Fantoni Urgo - Cirp Dictionary

17

Search using semantic structured informationExample – Patents (1)

Source: http://www.google.com/advanced_patent_search

Query: https://www.google.com/search?tbo=p&tbm=pts&hl=en&q=tube&num=10&gws_rd=ssl

Order By: Relevance Date: 18 August 2014

1. Cathode-ray tube amusement device, US2455992, 14 Dec 1948, Thomas T. Goldsmith Jr, Du Mont Allen B Lab Inc.

2. Draft tube for hydraulic turbine, US4515524, 7 May 1985, Richard K. Fisher Jr., Allis-Chalmers Corporation.

3. Vortex tube, US3173273, 16 Mar 1965, Fulton Charles D.

4. Twisted bourdon tube, US3463011, 26 Aug 1969, Werner Ries, Teldix Gmbh.

5. Eustachian tube stent, US6589286, 8 Jul 2003, Jason Litner.

Page 18: Fantoni Urgo - Cirp Dictionary

18

Search using semantic structured informationExample – Patents (2)

Source: http://www.google.com/advanced_patent_search

Query: https://www.google.com/search?tbo=p&tbm=pts&hl=en&q=tube+rolling+metal+forming&num=10&gws_rd=ssl

Order By: Relevance Date: 18 August 2014

1. Roller tube for awning and method of forming, US5383346, 24 Jan 1995, Ronald A. Laffler, White Consolidated Industries.

2. Apparatus for rolling threads into metal pipe, US2669139, 16 Feb 1954, Finch Harry J, Jones & Laughlin Steel Corp.

3. 半管滚压成型机 Half Pipe Roll Forming Machine, CN201613281U, 27 Oct 2010, 孙书霞 , 澳森凯(山东)机械制造有限公司 .

4. Mill for roll forming a fluted tube, EP0164233A2, 11 Dec 1985, Theodore H. Krengel, ALLIED TUBE & CONDUIT CORPORATION.

5. Calibration of an instrument for the cold-rolling of tubes, US6360575, 26 Mar 2002, Sergey Yurievich Zavodchikov, Joint Stock Company “Chepetskiy Mechanical Plant”.

Page 19: Fantoni Urgo - Cirp Dictionary

19

Search using semantic structured informationExample – Patents (3)

Source: http://www.google.com/advanced_patent_search

Query: https://www.google.com/search?tbo=p&tbm=pts&hl=en&q=tube+drawing+metal+forming&num=10&gws_rd=ssl

Order By: Relevance Date: 18 August 2014

1. Tube joint formed with adhesive and metal forming process, US5498096, 12 Mar 1996, James A. Johnson, Hoover Universal, Inc.

2. Composition and process for metal forming, US3454495, 8 Jul 1969, Horst Schneider, Hooker Chemical Corp.

3. Production method of internally-ribbed steel pipe and drawing plug for use therein, EP2228149A1, 15 Sep 2010, Kenichi Beppu, Sumitomo Metal Industries, Ltd.

4. Lubricant for use in non-chip metal forming, US4138348, 6 Feb 1979, Hans D. Grasshoff, Deutsche Texaco Aktiengesellschaft.

5. Production method of internally ribbed steel tube and drawing plug for use therein, US8281635, Kenichi Beppu, Sumitomo Metal Industries

Page 20: Fantoni Urgo - Cirp Dictionary

20

CIRP Terminology CommitteeAssets and Challenges

The information contained in the index of the CIRP Dictionary or the structure of the keywords can be used as a map to explore unknown sets of information (documents, patents, etc.).

Moreover, articles published and classified according to the STCs, constitute an homogeneous set of information that can be used to enrich the structured semantic data (e.g. the CIRP Dictionary) or as a “reference sample” to identify similarity among documents, research communities, researches, etc.

CIRP publications (CIRP Dictionary, CIRPedia, CIRP Annals, STC-related conferences) embed a layer of structured semantic information that constitutes an asset for the CIRP community.

Preserve, maintain, enhance and exploit these assets are important challenges the CIRP community will need to face in the future.