An Ontology-based Collaboration Recommender System using Patents

Sandra Geisler

, Rihan Hai

and Christoph Quix

Databases and Information Systems, RWTH Aachen University, Germany

Fraunhofer-Institute for Applied Information Technology FIT, St. Augustin, Germany

Keywords:

Ontology Engineering, Patent Analysis, Ontology Matching.

Abstract:

Successful research and development projects start with ﬁnding the right partners for the venture. Especially

for interdisciplinary projects, this is a difﬁcult task as experts from foreign domains are not known. Further-

more, the transfer of knowledge from research into practice is becoming more important in research projects

to enable the quick application of research results. This is in particular relevant for projects in medical engi-

neering. Patents and publications contain technical knowledge which can be exploited to ﬁnd suitable experts.

Patents are usually more product-oriented as the inventors have to describe an application area and products

might be protected by patents. On the other hand, scientiﬁc publications represent the state-of-the-art in re-

search. The challenge is ﬁnding the right mixture of research- or application-oriented experts from different

domains. Hence, we propose a recommender system for experts for a certain topic based on patent topic

clustering, ontologies, and ontology matching, which maps patents to corresponding innovation ﬁelds. The

medical engineering domain serves as a ﬁrst test bed, since projects in this area are highly interdisciplinary.

1 INTRODUCTION

Innovation drives research and industry. It is impor-

tant in both ﬁelds to be up-to-date to what will be

promising in the future. Especially, medical engi-

neering (ME) is an “innovative, strongly growing, and

promising industry in Germany”

. In ME, interdis-

ciplinary projects are very common as experts from

medicine, engineering, and other disciplines are re-

quired. Furthermore, this domain is highly dependent

on its innovative capabilities as product cycles are get-

ting shorter and shorter.

Ventures in research and industry rise and fall

with the expertise of the partners in the project team.

Hence, it is crucial for the success of innovative

projects and their proposals to ﬁnd suitable part-

ners. Studies demonstrate, that collaboration be-

tween research institutions and companies are beneﬁ-

cial for both, product and process innovations (Robin

and Schubert, 2013). Especially in interdisciplinary

projects, the search for experts in unfamiliar domains

is time consuming, cumbersome, and might not be as

successful as expected. Hence, to assist the process of

ﬁnding partners for a venture, a recommendation sys-

tem is desired which speeds up the search and helps

to discover collaboration opportunities.

http://www.bvmed.de/branchenbericht

Patents contain a wealth of technical information

used for the development of products, but are at the

same time hard to analyze as they are written using

special terminology (Aras et al., 2014; Zhang et al.,

2015). As patent inventors are not only experts in

their ﬁeld, but also have a product-oriented view on

ME research, they constitute interesting projects part-

ners. Therefore, we propose an approach using patent

clustering, ontology mappings, and ontology match-

ing to recommend collaboration opportunities.

In the mi-Mappa project

, we aim at ﬁnding suit-

able experts for ME projects based on patents and

innovation ﬁelds. According to (Schl

otelburg et al.,

2008), an innovation ﬁeld in ME is deﬁned as an area

which has signiﬁcant innovation activity, future po-

tential, and a value chain as complete as possible. The

main innovation ﬁelds for ME comprise (Schl

otelburg

et al., 2008; Deutsche Gesellschaft f

ur Biomed. Tech-

nik im VDE, 2012): Imaging Techniques, Prothe-

ses and Implants, Medical Information Systems and

Telemedicine, Interventional Devices, Systems, and

Techniques, In-vitro Technology, Special Therapy

and Diagnostic Systems, and cross-sectional topics,

such as patient safety.

In this paper, we propose an approach that com-

bines two complementary ways: 1. We build a proﬁle

http://www.dbis.rwth-aachen.de/mi-Mappa

Geisler, S., Hai, R. and Quix, C..

An Ontology-based Collaboration Recommender System using Patents.

In Proceedings of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2015) - Volume 2: KEOD, pages 389-394

ISBN: 978-989-758-158-8

389

of the expert, which includes her publications, web-

sites, and other business related information. This

comprises the identiﬁcation of the inventor with a cor-

responding author of scientiﬁc publications. If we

ﬁnd a corresponding author, we can match her pub-

lications to innovation ﬁelds to identify her special-

ization areas. We can use classiﬁcation terms present

in publication databases and semi-automatic ontology

matching to create the mappings. 2. If an inventor

could not be identiﬁed as an author of scientiﬁc pa-

pers, the patents are clustered by topic and these top-

ics are ﬁnally mapped to innovation ﬁelds using on-

tology matching.

In the following section, we will discuss existing

works in patent analysis and collaboration recommen-

dations. A full description of our approach is given in

Section 3. Finally, Section 4 concludes the paper.

2 RELATED WORK

Collaboration/Expert Recommender Systems.

The process of ﬁnding collaboration opportunities

often involves a manual process. For example, prede-

ﬁned criteria are analyzed, and scores are calculated

and weighted based on these criteria (Geum et al.,

2013; Awasthi et al., 2015). Partners in supply chains

can be found by using supervised and unsupervised

learning, statistics, and analyzing criteria (Wu and

Barnes, 2011). In the ﬁeld of ﬁnding partners

for R&D projects, no related semi-automatic or

automatic approach could be found.

Systems to ﬁnd experts for a certain topic are

based on self-disclosure (personal information main-

tained manually), authored documents, or social net-

work activity (Wang et al., 2013). The systems can

also be categorized into expert proﬁling and expert

ﬁnding (Balog and De Rijke, 2007). The most re-

cent works are using algorithms from social network

analysis, such as the link analysis algorithms PageR-

ank or HITS (Raﬁei and Kardan, 2015; Wang et al.,

2013) and graph-based algorithms (Rani et al., 2015).

We will concentrate on expert ﬁnding using authored

documents (e.g., patents & publications) as we do

not need (yet) a complete proﬁle of a researcher.

Many document-based Expert Recommender Sys-

tems (ERS) are only using enterprise-level documents

and are restricted to employees in the same company.

In contrast, we propose a document-based approach

which uses information from any publications and

patents available. The DEMOIR approach (Yimam-

Seid and Kobsa, 2003) also uses ontologies and do-

main models for expert ﬁnding, but they use them to

model the expertise only.

Patent Analysis using Ontologies. The usefulness

of ontologies has also been recognized for the patent

domain, especially for patent search (Bonino et al.,

2010). The PatExpert system, for example, uses a

network of ontologies and knowledge bases to enable

patent search, classiﬁcation, and clustering (Wanner

et al., 2008). Trappey et al. propose a system that cal-

culates the conditional probability that, given a spe-

ciﬁc text chunk is present in the document, the chunk

is mapped to a speciﬁc concept of a given ontol-

ogy (Trappey et al., 2009). Patent similarity is then

based on the number of common matched concepts.

This approach restricts the clustering to the terms of

the ontology which might lead to missing important

terms not present in the ontology.

Patent Clustering. An overview of patent docu-

ment contents can be retrieved by clustering. Tseng et

al. propose a full-text patent clustering methodology

which includes document clustering, term clustering,

and multi-stage clustering to avoid skewed distribu-

tion among clusters (Tseng et al., 2007). TF or IDF

(see section 3.2) ﬁltered terms are clustered accord-

ing to their co-occurrence. Moreover, each cluster

obtains a summary title by statically calculating the

most frequent terms in the clusters with correlation

coefﬁcient method. A bibliometric approach based on

co-citation analysis is introduced in (Mogee and Ko-

lar, 1999). The co-cited documents are linked under

the assumption that they share the subject matter. The

result of the approach also indicates core competen-

cies in the corresponding industrial ﬁeld. However,

using co-citation to group patents may lead to super-

ﬁcial results due to the lack of internal knowledge of

the patents (Yoon and Park, 2004). Another drawback

is that patents without references are excluded from

this approach. Trappey et al. describe a methodol-

ogy to cluster patents in three steps. First they extract

the key phrases of a patent, i.e., they use an ontology-

based, statistical method to extract key phrases which

represent an important topic in the document. After-

wards they build Technology Clusters of these key

phrases using an non-exhaustive overlapping cluster

algorithm proposed by Chen and Hu (Trappey et al.,

2010; Chen and Hu, 2006). By calculating the cumu-

lative weights for the key phrases of a document, they

can determine the Technology Cluster for a patent. In

the last step, they use the same clustering technique to

cluster the patent documents.

In summary, all of the approaches may cover a

part of our approach, but we present a novel approach

which combines the use of patent analysis, cluster-

ing, ontology design, and ontology matching to rec-

ommend experts for a R&D collaboration.

KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development

390

Figure 1: Architecture to map patents to innovation ﬁelds.

3 APPROACH

Based on the processed information type, patent anal-

ysis can be categorized into quantitative and qualita-

tive approaches

, which we both utilize in our ap-

proach. Qualitative approaches extract patent meta-

data, e.g., inventors, references, afﬁliation, while

quantitative methods process the full-text, such as

claims and abstracts of patent documents. The goal

of the overall approach, depicted in Figure 1, is to

map inventors of a patent and subsequently the cor-

responding patents to innovation ﬁelds. We propose

to do this in two different, but supplementary ways:

1. Match inventors with publication authors. 2. Clus-

ter patents based on the topics they cover. For both

approaches, we preselect patents based on a search

with keywords extracted from the description of the

intended project. The resulting patents can be used

to initialize the clustering. In both approaches the re-

sults (publications or topics) will be matched to ex-

isting medical ontologies. If a mapping between the

matched medical terms and concepts of an innovation

ﬁeld exists, a link between patent inventor and inno-

vation ﬁelds has been found.

In the following, we detail the two process ﬂows.

Subsequently, we describe the design of the utilized

ontologies and the matching in the ontology network.

3.1 Inventor-Author-Identiﬁcation

It is ﬁrst assumed, that an inventor may be also work-

ing as a researcher and is publishing scientiﬁc arti-

cles in the same domain. We map inventors to au-

thors, because articles published in journals or con-

ferences are often classiﬁed by publishers and pub-

lication databases. Those classiﬁcations usually use

terms which are easier to harvest and match to do-

main ontologies than the abstract International Patent

Classiﬁcation (IPC) or similar classiﬁcations. Very

promising are bibliographic search engines such as

http://www.wipo.int/sme/en/documents/

patent information fulltext.html

Web of Science, or PubMed, which offer classiﬁca-

tion of papers according to well known medical tax-

onomies. We will search the bibliographic databases

by author name using a corresponding API. For au-

thor identiﬁcation, we use a multi-step process that

uses clustering techniques and statistics to determine

the highest probability of an author being the same

person as an inventor of a certain patent. We extract

the keywords and classiﬁcation terms from the meta-

data of the papers and match them with medical on-

tologies which in turn will be mapped to an innovation

ﬁeld ontology.

3.2 Topic Clustering

A study showed that 70-90% of technological knowl-

edge is only published in patents

. Hence, it can

be assumed that there exist inventors who have only

published patents and an alternative way of mapping

patents and inventors to an innovation ﬁeld has to be

found for those persons. We propose to use topic clus-

tering of the patents to do so. This second approach

can also be used to verify the results of the ﬁrst ap-

proach.

In patent analysis, classiﬁcations such as the

IPC are often too broad for speciﬁc analytical us-

age (Tseng et al., 2007) and more detail-oriented cat-

egorizations are needed. Clustering methods group

objects such that the similarity between objects in the

same cluster is greater than the similarity of objects in

different clusters. The similarity is usually measured

in terms of their relative position in an n-dimensional

space using Euclidean or Manhattan distances.

We utilize a set of common preprocessing tech-

niques to facilitate feature extraction, indexing, and

clustering. These comprise, amongst others, docu-

ment parsing, tag removal, tokenization, and lower-

casing. Additional steps, such as stemming, pruning,

and stopword removal, help to reduce the term set size

and increase its quality, improving clustering accu-

racy (Gonc¸alves et al., 2010). In each document, only

the key terms are selected to present the features of

the document, utilizing Inverse Document Frequency

(IDF) and Term Frequency (TF) within certain thresh-

olds. These terms are weighed based on TF ˆ IDF,

followed by calculating the similarity using cluster-

ing algorithms. Documents are usually merged to

clusters successively (hierarchical clustering) or dis-

tributed to certain clusters deﬁned in the beginning

(partition clustering). We use both kinds of distance-

based clustering in our approach, including K-means

http://www.integrityip.com/Patent Library/

Community/Other/GlobalPatentSources.pdf

An Ontology-based Collaboration Recommender System using Patents

391

and K-medoid algorithms with pre-chosen centroids

from the query results.

Moreover, it is common in patent analysis that a

patent includes multiple features, claims, or inventors.

Hence, the non-exhaustive overlapping clustering al-

gorithm (Trappey et al., 2010) is adopted in our ap-

proach. Finally, each cluster receives a title generated

based on the top k frequent terms (Yang et al., 2000).

Furthermore, as performance is an issue in full text

analysis of a large document collection, we apply the

text analysis only to a part of each patent document

(e.g., the ﬁrst part of the abstract, claims, or introduc-

tion). It has been proven that such an approach may

achieve better performance than using full texts (Fall

et al., 2003).

3.3 Ontologies

Selection of Existing Domain Ontologies. Our ap-

proach is heavily relying on the mappings to medical

ontologies and subsequently from medical ontologies

to the innovation ﬁeld ontology. A plethora of medi-

cal ontologies exist. Hence, we have to analyze which

set of ontologies covers as many terms as possible,

describing the innovation ﬁelds.

We made a ﬁrst analysis by searching for ontolo-

gies in the Bioportal

search engine using terms

describing innovation ﬁelds. The Bioportal search

engine is the most comfortable and comprehensive

search engine in the life science domain. In addi-

tion, it offers several useful tools, e.g., an ontology

recommendation tool based on keywords or full-texts.

Moreover, we used the Ontology Lookup Service

and the Ontobee

search engine to have a broad

overview. For the search, 174 terms from the six inno-

vation ﬁelds extracted from the reports (Schl

otelburg

et al., 2008; Deutsche Gesellschaft f

ur Biomed.

Technik im VDE, 2012) have been used. The

most promising four ontologies found were the Na-

tional Cancer Institute (NCIT) Thesaurus, the Sys-

tematized Nomenclature of Medicine - Clinical Terms

(SNOMEDCT), MeSH, and the Robert Hoehndorf

Version of MeSH (RHMeSH). For these we did a cov-

erage analysis presented in Figure 2. The coverage is

the percentage of the innovation ﬁeld terms present in

each of the ontologies.

Note that no ontology really outperforms the oth-

ers and that the overall coverage is very low. Hence,

we decided to analyze the coverage by adding one

ontology after another, to see the gain of adding fur-

ther ontologies. We used the most promising ontolo-

http://bioportal.bioontology.org

http://www.ebi.ac.uk/ontology-lookup

http://www.ontobee.org

Imaging

Techniques

Prostheses &

Implants

Telemedicine Operative &

Interventional

Dev. and Sys.

In-Vitro

Diagnostics

Special

Therapies &

Diagnosis Sys.

Complete

NCIT SNOMEDCT MeSH RHMeSH

Figure 2: Coverage of search terms in selected ontologies.

Imaging

Techniques

Prostheses &

Implants

Telemedicine Operative &

Interventional

Dev. and Sys.

In-Vitro

Diagnostics

Special

Therapies &

Diagnosis Sys.

Complete

NCIT NCIT + MeSH NCIT + MeSH + SNOMEDCT NCIT + MSH + SNOMEDCT + RHMeSH

Figure 3: Coverage based on combination of ontologies.

Figure 4: The Bioportal Recommender Tool.

gies identiﬁed before and started with the NCI The-

saurus. Figure 3 shows the results. It can be noted,

that we gain about 10% coverage using all ontolo-

gies. The biggest gain is achieved after adding the

MeSH ontology. An analysis with the same terms us-

ing the Bioportal Recommender tool as depicted in

Figure 4 delivers a similar result. The Recommender

tool analyzes the annotations which can be found us-

ing the given terms. The coverage is calculated tak-

ing into account amongst others the mappings to and

synonyms from other ontologies, and the size of the

ontologies (Jonquet et al., 2010). Hence, it is not di-

rectly comparable to the manually created coverage

result described above. There are two recommended

ontology sets ranked highest: The ﬁrst comprises the

NCIT, MeSH, and the Computer Retrieval of Infor-

mation on Scientiﬁc Projects (CRISP) and resulted in

a coverage of 83.6% and an overall result of 68.9%.

The second is congruent with our selection: NCIT,

MeSH, and SNOMEDCT and resulted in a coverage

KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development

392

of 84.6% and an overall score of 68.9%. Further-

more, all the ontologies have to be assessed accord-

ing to their quality. A detailed quality analysis will be

made using acknowledged quality criteria (Vrande

2009; Gomez-Perez, 2004). We are still in the process

of ﬁnding a suitable tool, such as the Ontology Pit-

fall Scanner (OOPS!)

(Poveda-Villal

on et al., 2012),

which will assist us in this regard.

Requirements Analysis & Design of the Innovation

Field Ontology. To reach our goal to map publica-

tions and clusters to innovation ﬁelds, we also need

a new ontology which only represents the innova-

tion ﬁelds and important terms describing them. To

that end, we will make a detailed requirements anal-

ysis including interviews with domain experts, anal-

ysis of existing ontologies, and an intensive litera-

ture research. Where applicable, we will stick to the

NeOn methodology (Su

arez-Figueroa, 2010) and es-

pecially for the requirements analysis, the creation

of an Ontology Requirements Speciﬁcation Docu-

ment (Su

arez-Figueroa et al., 2009) will be useful. In

a ﬁrst step, we have already extracted a preliminary

selection of 174 terms (the same terms as for ontology

search), which corresponds to the scenario “Reusing

and reengineering non-ontological resources” of the

NeOn methodology. The terms can be used to make

a ﬁrst draft of a preliminary ontology which is veri-

ﬁed during an expert interview. The ontology design

will be accompanied by its evaluation using the crite-

ria and tools mentioned above.

3.4 Ontology Matching & Mappings

To identify which collaboration partner is working in

which innovation ﬁeld, we have to integrate all the in-

formation we gathered so far. We need to map the

cluster terms and the publications, respectively, to the

domain ontologies. The mappings between the do-

main ontologies and the innovation ﬁeld ontology can

be established during design time, because it is not

expected that they will change frequently. Ontology

matching systems (e.g., our tool GeRoMeSuite (Ken-

sche et al., 2007)) will be used to identify a ﬁrst set

of mappings. Additionally, existing mappings created

by the BioPortal can be used to infer further mappings

to the innovation ﬁeld ontology.

Bioportal also provides prepared mappings be-

tween ontologies. It creates the mappings either using

the LOOM algorithm, the Uniﬁed Medical Language

System (UMLS) concept unique identiﬁers (CUI),

and the Open Biological and using Biomedical On-

http://oops.linkeddata.es

tologies (OBO) xref properties

(Ghazvinian et al.,

2009). These mappings can be retrieved via a REST

API offered by the BioPortal website. Afterwards, a

domain expert veriﬁes the detected mappings.

If necessary, the creation of mappings between

publication classiﬁcations and domain ontologies can

be prepared during design time. We will also use

the semi-automatic matching process described be-

fore. More challenging is the creation of mappings

between the clusters and the domain ontologies. This

has to be done during run time, as the cluster terms

are not known in advance. We plan to use also the

matching algorithms provided by GeRoMeSuite for

this step.

4 CONCLUSION

We have presented an innovative ontology-based

approach for recommending experts for research

projects in ME. We are making extensive use of ontol-

ogy engineering in our approach, e.g., analysis, cre-

ation, and matching of ontologies, deﬁning require-

ments for ontologies, and evaluation of ontologies.

Also techniques from other areas, such as text min-

ing and patent analysis, are included in our approach.

Our current work focuses on the modeling of the on-

tologies and the selection of the clustering methods.

Performance is an issue for text clustering, as we want

to have an interactive system.

The work is still in an early stage and we have to

see how the integration of text clustering, topic mod-

eling, patent analysis, and ontology matching per-

forms. The various techniques are also challenging

tasks if they are considered separately, but the com-

bination of the techniques may show an innovative

method for exploring unknown research ﬁelds. Our

approach is not limited to the ﬁeld of ME; however,

the availability of a huge number of ontologies in the

life sciences contributes to our approach.

ACKNOWLEDGEMENTS

This work has been supported by the Klaus Tschira

Stiftung gGmbH in the context of the mi-Mappa

project (http://www.dbis.rwth-aachen.de/mi-Mappa/,

project no. 00.263.2015). We would like to thank

our partners in the mi-Mappa project for their fruitful

ideas. We would further like to thank Tanja Schmelter

for her work on ontology analysis.

http://www.bioontology.org/wiki

An Ontology-based Collaboration Recommender System using Patents

393

REFERENCES

Aras, H., Hackl-Sommer, R., Schwantner, M., and Sofean,

M. (2014). Applications and challenges of text mining

with patents. In Proc. Intl. Workshop on Patent Mining

and its Applications. Stiftung Univ. Hildesheim.

Awasthi, A., Adetiloye, T., and Crainic, T. G. (2015). Col-

laboration partner selection for city logistics planning

under municipal freight regulations. Applied Mathe-

matical Modelling.

Balog, K. and De Rijke, M. (2007). Determining expert pro-

ﬁles (with an application to expert ﬁnding). In IJCAI,

volume 7, pages 2657–2662.

Bonino, D., Ciaramella, A., and Corno, F. (2010). Re-

view of the state-of-the-art in patent information and

forthcoming evolutions in intelligent patent informat-

ics. World Patent Information, 32(1):30–38.

Chen, Y.-L. and Hu, H.-L. (2006). An overlapping cluster

algorithm to provide non-exhaustive clustering. Eu-

rop. J. of Operational Research, 173(3):762–780.

Deutsche Gesellschaft f

ur Biomed. Technik im VDE

(2012). Empfehlungen zur Verbesserung der In-

novationsrahmenbedingungen f

ur Hochtechnologie-

Medizin. Technical report, VDE.

Fall, C. J., T

orcsv

ari, A., Benzineb, K., and Karetka, G.

(2003). Automated categorization in the international

patent classiﬁcation. In ACM SIGIR Forum, vol-

ume 37, pages 10–25. ACM.

Geum, Y., Lee, S., Yoon, B., and Park, Y. (2013). Identify-

ing and evaluating strategic partners for collaborative

r&d: Index-based approach using patents and publi-

cations. Technovation, 33(6):211–224.

Ghazvinian, A., Noy, N., and Musen, M. (2009). Creating

mappings for ontologies in biomedicine: simple meth-

ods work. In AMIA Ann. Symp. Proc., pages 198–202.

Gomez-Perez, A. (2004). Ontology evaluation. In Staab,

S. and Studer, R., editors, Handbook on Ontologies,

pages 250–273. Springer.

Gonc¸alves, C. A., Gonc¸alves, C. T., Camacho, R., and

Oliveira, E. C. (2010). The impact of pre-processing

on the classiﬁcation of medline documents. In PRIS,

pages 53–61.

Jonquet, C., Musen, M. A., and Shah, N. H. (2010). Build-

ing a biomedical ontology recommender web service.

J. Biomedical Semantics, 1(S-1):S1.

Kensche, D., Quix, C., Li, X., and Li, Y. (2007). GeRoMe-

Suite: A system for holistic generic model manage-

ment. In Proc. VLDB, pages 1322–1325.

Mogee, M. E. and Kolar, R. G. (1999). Patent co-citation

analysis of eli lilly & co. patents. Expert Opinion on

Therapeutic Patents, 9(3):291–305.

Poveda-Villal

on, M., Su

arez-Figueroa, M. C., and G

omez-

erez, A. (2012). Validating ontologies with oops!

In Knowledge Engineering and Knowledge Manage-

ment, pages 267–281. Springer.

Raﬁei, M. and Kardan, A. A. (2015). A novel method for

expert ﬁnding in online communities based on concept

map and pagerank. Human-centric Computing and

Information Sciences, 5(1):1–18.

Rani, S. K., Raju, K., and Kumari, V. V. (2015). Expert

ﬁnding system using latent effort ranking in academic

social networks. Intl. J. of Information Technology

and Computer Science, 2:21–27.

Robin, S. and Schubert, T. (2013). Cooperation with public

research institutions and success in innovation: Ev-

idence from france and germany. Research Policy,

42(1):149–166.

Schl

otelburg, C., Weiß, C., Hahn, P., Becks, T., and

uhlbacher, A. C. (2008). Identiﬁzierung von Innova-

tionsh

urden in der Medizintechnik. Technical report,

Bundesministeriums f

ur Bildung und Forschung.

arez-Figueroa, M. C. (2010). NeOn Methodology for

building ontology networks: speciﬁcation, schedul-

ing and reuse. PhD thesis, Universidad Politecnica

de Madrid.

arez-Figueroa, M. C., G

omez-P

erez, A., and Villaz

on-

Terrazas, B. (2009). How to write and use the on-

tology requirements speciﬁcation document. In Proc.

OTM 2009, pages 966–982. Springer.

Trappey, A. J., Trappey, C. V., Hsu, F.-C., and Hsiao, D. W.

(2009). A fuzzy ontological knowledge document

clustering methodology. IEEE Trans. on Systems,

Man, and Cybernetics, Part B, 39(3):806–814.

Trappey, C. V., Trappey, A. J., and Wu, C.-Y. (2010). Clus-

tering patents using non-exhaustive overlaps. System

Science and System Engineering, 19(2):162–181.

Tseng, Y.-H., Lin, C.-J., and Lin, Y.-I. (2007). Text mining

techniques for patent analysis. Information Process-

ing & Management, 43(5):1216–1247.

Vrande

c, D. (2009). Ontology evaluation. In Staab, S. and

Studer, R., editors, Handbook on Ontologies, chap-

ter 13, pages 293–313. Springer.

Wang, G. A., Jiao, J., Abrahams, A. S., Fan, W., and Zhang,

Z. (2013). Expertrank: A topic-aware expert ﬁnding

algorithm for online knowledge communities. Deci-

sion Support Systems, 54(3):1442–1451.

Wanner, L., Baeza-Yates, R., Br

ugmann, S., Codina, J., Di-

allo, B., Escorsa, E., Giereth, M., Kompatsiaris, Y.,

Papadopoulos, S., Pianta, E., et al. (2008). Towards

content-oriented patent document processing. World

Patent Information, 30(1):21–33.

Wu, C. and Barnes, D. (2011). A literature review of

decision-making models and approaches for partner

selection in agile supply chains. Purchasing and Sup-

ply Management, 17(4):256–274.

Yang, Y., Ault, T., Pierce, T., and Lattimer, C. W.

(2000). Improving text categorization methods for

event tracking. In Proc. of the 23rd Intl. Annual ACM

SIGIR Conf., pages 65–72. ACM.

Yimam-Seid, D. and Kobsa, A. (2003). Expert-ﬁnding sys-

tems for organizations: Problem and domain analysis

and the demoir approach. J. of Organizational Com-

puting and Electronic Commerce, 13(1):1–24.

Yoon, B. and Park, Y. (2004). A text-mining-based patent

network: Analytical tool for high-technology trend.

The Journal of High Technology Management Re-

search, 15(1):37–50.

Zhang, L., Li, L., and Li, T. (2015). Patent mining: A sur-

vey. ACM SIGKDD Expl. Newsletter, 16(2):1–19.

KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development

394