=Paper=
{{Paper
|id=Vol-2218/paper4
|storemode=property
|title=WordNet based Semantic Similarity Measures for Process Model Matching
|pdfUrl=https://ceur-ws.org/Vol-2218/paper4.pdf
|volume=Vol-2218
|authors=Khurram Shahzad,Ifrah Pervaz,Adeel Nawab
|dblpUrl=https://dblp.org/rec/conf/bir/ShahzadPN18
}}
==WordNet based Semantic Similarity Measures for Process Model Matching==
https://ceur-ws.org/Vol-2218/paper4.pdf
WordNet-based Semantic Similarity Measures for
Process Model Matching
Khurram Shahzad, Ifrah Pervaz, Rao Muhammad Adeel Nawab*
Punjab University College of Information Technology, University of the Punjab, Lahore.
*
Department of Computer Science, COMSATS Institute of Information Technology, Lahore.
khurram|ifrah@pucit.edu.pk, adeelnawab@ciitlahore.edu.pk
Abstract. Process Model Matching (PMM) refers to the automatic identification
of corresponding activities between a pair of process models. Due to the wider
applicability of PMM techniques several semantic matching techniques have
been proposed. However, these techniques focus on utilizing few word-to-word
(word-level) similarity measures, without giving due consideration to activity-
level aggregation methods. The inadequate attention to the choice of activity-
level methods limit the effectiveness of the matching techniques. Furthermore,
there are some WordNet-based semantic similarity measures that have shown
promising results for various text matching tasks. However, the effectiveness of
these measures has never been evaluated in the context of PMM. To that end, in
this paper we have used five word-level semantic similarity measures and three
sentence-level aggregation methods to experimentally evaluate the effectiveness
of their 15 combinations for PMM. The experiments are performed on the three
widely used PMMCโ15 datasets. From the results we conclude that, a) Jiang sim-
ilarity is more suitable than the mostly used Lin similarity, and b) QAP is the
most suitable sentence-level aggregation method.
Keywords: Business Process Models, Process Model Matching, Semantic Sim-
ilarity, WordNet-based similarity measures.
1 Introduction
Business process models are the conceptual models that explicitly represent the busi-
ness operations of an enterprise. These models are widely accepted as a useful resource
for a variety of purposes ranging from representing requirements for software develop-
ment to configuring ERP systems. Process Model Matching (PMM) refers to identify-
ing the activities between two process models that represent similar or identical func-
tionality [1]. A pair of activities that represent similar or identical functionality is called
a corresponding pair and the involved activities are called corresponding activities [2].
Figure 1 shows the example process models of two universities, University A and Uni-
versity B, and correspondences between their activities. In the figure, each correspond-
ence between a pair of activities is marked by a shaded area.
An accurate identification of corresponding activities is of higher significance for
the BPM community due to its widespread application areas, such as identifying clones
of process models, searching process models and harmonizing process models [3]. To
that end, a plethora of automatic techniques have been proposed [4]. Despite the exist-
ence of several matching techniques, the need for enhancing the accuracy of matching
techniques has been widely pronounced during the recent years [4, 5]. For instance, a
comprehensive survey of the state-of-the-art has made imperative revelations about
process model matching techniques [5]. The two notable ones are, 1) 21 out of 35 tech-
niques use the most basic syntactic measures, and 2) Lin similarity is the most promi-
nent semantic similarity measure.
Fig. 1. Illustration of process model matching
In this study we contend that there are several word-level semantic similarity
measures that have shown promising results for various text processing tasks [6, 7].
However, an empirical assessment of these competing measures has never been con-
ducted in the context of process model matching. Consequently, a well-grounded rec-
ommendation about the choice of a semantic similarity measure is non-existent. Fur-
thermore, the presently used similarity measures merely focus on the word-to-word se-
mantic similarity, without paying adequate attention to the aggregation of word-level
similarity scores to an activity-level similarity score. This arbitrary selection of sen-
tence-level aggregation method, such as average score, may impede the effectiveness
of the matching techniques. To that end, in this study we evaluate the effectiveness of
five WordNet-based word-to-word semantic similarity measures and three sentence-
level methods, which extend word-level semantic similarity scores to an activity-level
similarity score. The effectiveness of all (fifteen) combinations of five word-level se-
mantic similarity measures and three sentence-level methods are evaluated using the
three PMMCโ15 datasets.
The rest of the paper is organized as follows: Section 2 provides an overview of the
word-level and sentence-level semantic similarity measures. Section 3 and 4 presents
the experimental setup and results of the experiments, respectively. Section 5 provides
an overview of the related work. Finally, Section 6 concludes the paper.
34
2 WordNet-based Semantic Similarity Measures
The semantic similarity methods that we have applied for identifying corresponding
activities between process models are based on WordNet. WordNet is widely acknowl-
edged as a valuable source to find semantic similarity between two words as it organizes
words based on lexical relations and then defines semantic relations between those lex-
ically related synsets. The lexical relationships are categorized into two subcategories:
synsets, and antonyms, whereas semantic relations are categorized into five sub cate-
gories: hyponyms, meronyms, co-ordinate terms, entailment of a verb, and troponym
of a verb. Synsets are related with other synsets to form a hierarchical structure of con-
ceptual relations. In the WordNet version 2.0, there are nine noun hierarchies that in-
clude 80,000 concepts and 554 verb hierarchies that are made up of 13,500 concepts.
All the concepts are linked to a unique root, called entity.
2.1 Word-level Semantic Similarity Measures
We have selected five well established and widely used word-level semantic similarity
measures to compute the degree of semantic similarity between activity pairs. These
are: Resnik similarity [7], Jiang similarity [8], Leacock similarity [6], Lin similarity [9],
and Wu similarity [10]. The methods have been previously used for lexical and textual
semantic relatedness [11, 13], word sense disambiguation [12], gene and sequence
matching [14], generating sentences from pictures [15], paraphrasing [16], sentiment
analysis [17] and topic modeling [18, 19]. A brief overview of these measures is as
follows:
Resnik Similarity. Resnik similarity relies on is-a relationship in the WordNet taxon-
omy, where each node represents a unique WordNet synset or concept. According to
this measure two nodes are considered more similar if they share more information.
This shared information is specified by Information Content (IC) of the nodes that sub-
sumes these nodes in a taxonomy. Formally, IC is calculated as follows:
๐ผ๐ถ = โ ๐๐๐ ๐(๐ถ)
Let C1 and C2 be the two concept nodes in WordNet taxonomy and concept node C
is the lowest common subsumer node of nodes C1 and C2. Furthermore, let P(C) is the
probability of occurrence of longest common subsumer node C and probability of node
C is simply found by normalizing occurrences of concepts with total number of nouns
in the taxonomy.
( )
๐(๐ถ) = ๐๐๐ ๐(๐) = โ โ ( ) ๐๐๐ข๐๐ก(๐)
Where, W(C) is the set of concepts in which word w occurs and each occurrence of
a word is considered as occurrence of all concepts containing that word. The Resnik
similarity is referred as maximal IC over all concepts to which both words belong. For-
mally, it is defined as follows:
๐ ๐๐ ( ๐ถ1, ๐ถ2) = ๐ผ๐ถ (๐ฟ๐ถ๐ ( ๐ถ1, ๐ถ2))
Where, LCS is the lowest common subsumer of concept nodes C1 and C2 defined
as the common parent of these with minimum node distance.
35
Jiang Similarity. This method uses corpus statistical information i.e. Information Con-
tent (IC) and nodes path in is-a taxonomy for computing similarity, where, IC is meas-
ure of occurrence of the concept in the corpus. Given a word pair C1 and C2, this meas-
ure computes similarity between the words by using following equation:
๐ ๐๐ ( ๐ถ1, ๐ถ2) = ( ) ( )
โ ( )
Where, IC stands for information content and LCS is the Lowest Common Subsumer
of concepts C1 and C2 defined as the common parent of these with minimum node
distance.
Leacock Similarity. This similarity measure is based on a node based approach using
is-a taxonomy in the WordNet. When considering the WordNet taxonomy, each node
represents a unique concept (or synset) in the taxonomy. Subsequently, the degree of
similarity between a word pair is computed by calculating the shortest path between
two concepts (represented as nodes), and dividing it by twice the maximum depth of
the graph. Formally, it is represented as follows:
๐ ๐๐ ( ๐ถ1, ๐ถ2) = โ๐๐๐
โ
In the equation, C1 and C2 are the two concepts represented by nodes, shortest path
length is the minimum path length from node C1 to node C2 by using node counting,
and depth is the number of maximum nodes from root node to a leaf node.
Lin Similarity. According to this measure, similarity between two concepts is ex-
pressed as the similarity between generic terms belonging to these concept classes, ra-
ther than measuring similarity between all terms. For instance, word โdesignโ belongs
to the concept class โblueprintโ, and the word โconstructโ belongs to the concept class
named โconceptโ. According to Lin, the similarity between these words should be the
same as the similarity between two synsets โblueprintโ and โconceptโ to which these
words belong. Formally, if there is a word x1โ C1 and a word x2โ C2 the information
shared by two words can be expressed by C, the most specific class that subsumes both.
The similarity is then computed as:
โ ( )
๐ ๐๐ (๐ฅ , ๐ฅ ) =
( ) ( )
Where C is the most specific class that subsumes concepts C1 and C2 is the common
parent of the two concepts with a minimum node distance. log P (C), log P (C1) and
log P (C2) are log likelihood of the occurrence of concepts C, C1 and C2.
Wu Similarity. This similarity measure relies on the depth of both concept nodes, and
depth of lowest common subsumer. The similarity between two concepts is then com-
puted by using the following Equation.
โ ( )
๐ ๐๐ (๐ถ1, ๐ถ2) =
( ) ( )
Where LCS is the lowest common subsumer of concepts C1 and C2 defined as the
common parent of C1 and C2 with minimum node distance. Depth (C1) represents the
number of nodes from C1 to LCS node C, depth (C2) represents the number of nodes
36
from C2 to LCS node C and depth (LCS) represents the number of nodes from LCS
node C to root node.
2.2 Sentence-Level Similarity Methods
The preceding section presented various WordNet-based semantic similarity measures
for computing word-level similarity. These measures compute similarity between a pair
of words, however, PMM refers to computing similarity between activity pairs. There-
fore, there is a need to combine the word-level measures with sentence-level methods,
where each label is considered as a sentence. For this purpose, we applied three methods
which extend word-level measures to sentence-level methods. These methods are
Greedy Pairing [21], Optimal Matching [22], and Quadratic Assignment Problem
(QAP) [23]. A brief overview of each method is described below.
Greedy Pairing. Using this method, at first both sentences are tokenized. After that,
word-level semantic similarity method is used to search the maximum semantic simi-
larity of each token in the first sentence with all the tokens in the second sentence.
These maximum similarities are weighted using Inverse Document Frequency (IDF)
scores. Maximum similarity scores of all the tokens in the first sentence are computed,
summed up and resulting score is normalized with maximum sentence length. The same
steps are repeated to find the maximum mappings for each token of the second sentence
with the first sentence. The final similarity score between sentence pair is obtained by
computing the average scores obtained using sentence one and two.
Optimal Matching. This method is based on combinatorial matching problem, where
for a given weighted bipartite graph the problem is to find maximum matching of graph.
Bipartite graph is the graph whose nodes can be divided into two disjoint sets. Using
this approach, the two sentences S1 and S2 are considered part of a weighted bipartite
graph G = S1 โช S2, where words in sentences are represented as nodes of graph and
the weight of edges between these nodes corresponds to similarity score between the
respected nodes. The task is to select those node pairs in matching M such that overall
sum of all selected node pair is maximum.
Quadratic Assignment Problem (QAP). This approach finds an optimal assignment
of word in first sentence to second sentence using word-level similarity measure and at
the same time maximizes the similarity between syntactic dependencies of words pair.
The Koopmans-Beckmann formulation of the QAP problem is used. The goal is to
maximize the objective function QAP (F, D, B) where F and D captures syntactic de-
pendencies between words in two sentences respectively and B captures the word-to-
word similarity across the two sentences.
37
3 Experimental Setup
For the experiments we have used three well established datasets, developed by experts
and used in Process Model Matching Contest 2015 (PMMCโ15). Since the competition,
the datasets are widely used for the evaluation of process model matching techniques
[24]. The datasets are named as, University Admissions (UA), Birth Registration (BR),
and Asset Management (AM) datasets. Below, we present a brief overview of the three
datasets. The UA dataset is composed of 9 process models about admission to nine
German universities and 36 pairs of process models. In addition to that, the dataset
includes gold standard correspondences between equivalent activities. The specifica-
tions of the three datasets are given below in Table 1.
The BR dataset includes 9 process models, 36 pairs of these nine models and gold
standard correspondences. The models represent birth registration process of different
countries: Germany, Russia, South Africa, and the Netherlands. The collection includes
both 1:1 and 1: n correspondences. The AM dataset consists of 36 process model pairs
selected from 72 process models of the SAP reference model collection [24]. The se-
lected process models cover different aspects from the area of asset management.
Table 1. Specifications of the collected PMMCโ15 datasets.
UA dataset BR dataset AM dataset
No of Activities (Min) 12 9 1
No of Activities (Max) 45 25 43
No of Activities (Avg) 24.2 17.9 18.6
No of 1:1 Correspondences 268 156 140
No of 1:n Correspondences 360 427 82
4 Results and Analysis
This section presents the analysis of the results, which are obtained by applying five
combinations of five word-level semantic similarity measures and 3 sentence-level
methods.
The main goal of experiments is to classify each activity pair as โequivalentโ or โnon-
equivalentโ. Since, the semantic similarity methods used in this study return a numeric
score between 0 and 1, we have converted these numeric scores into binary 0 (non-
equivalent) and 1 (equivalent) at nine different thresholds from 0.1 to 0.9 with a gap of
0.1. However, due to space limitations we have used a cut-off threshold 0.7 because
multiple matching systems participating in the latest episode of the Process Model
Matching Contest 2015 achieved promising results at this threshold. This threshold
value represents that each activity pair for which the similarity score is greater than or
equal to 0.7 is marked as equivalent (or 1) and non-equivalent (or 0) otherwise. The F1
scores at 0.7 threshold are presented in Table 2 and the remaining results are made
available for download. For each dataset, the word-level measure that obtained the
highest F1 score for a sentence-level method is highlighted in bold. Therefore, each
38
sentence-level method has at least one bold value. Furthermore, we have underlined the
word-level measure that obtained the highest F1 score for a dataset, independent of any
sentence-level aggregation method.
Table 2. Results of all the techniques for PMMCโ15 datasets.
Sentence Level Word-level UA Dataset BR Dataset AM Dataset
Resnik 0.516 0.532 0.464
Jiang 0.516 0.534 0.464
Greedy 0.453
Leacock 0.487 0.509
Pair
Lin 0.513 0.533 0.455
Wu 0.495 0.516 0.456
Resnik 0.519 0.532 0.464
Jiang 0.516 0.534 0.464
Optimal 0.454
Leacock 0.489 0.516
Pairing
Lin 0.520 0.533 0.456
Wu 0.495 0.525 0.457
Resnik 0.520 0.532 0.464
Jiang 0.520 0.534 0.464
QAP Leacock 0.498 0.522 0.455
Lin 0.525 0.534 0.457
Wu 0.508 0.525 0.459
A brief analysis of the results is as follows.
Difficulty level of datasets. From the table it can be observed that there is a clear differ-
ence between the performance of all techniques for the three datasets. That is, all com-
binations of techniques obtained high F1 scores for UA dataset, moderate F1 scores for
BR dataset, and low F1 scores for AM dataset. These results indicate that the corre-
sponding activity pairs of the AM dataset are harder-to-detect than that of UA and BR
datasets. Furthermore, the corresponding activities in the BR datasets are harder-to-
detect than that of the UA dataset.
Performance variation across word-level measures. From Table 2 it can be observed
that in the case of Greedy pairing sentence-level method, Jiang similarity obtained the
highest F1 scores for all the three datasets (0.516, 0.534 and 0.464). Furthermore, for
Optimal pairing sentence-level method, Jiang similarity obtained the highest F1 scores
for two datasets, BR and AM datasets (0.534 and 0.464) whereas, Lin similarity ob-
tained the highest F1 score for one dataset, UA dataset. Similarly, for QAP pairing,
Jiang similarity obtained the highest F1 scores for two datasets, BR and AM datasets,
whereas Lin similarity obtained the highest F1 score for one dataset, UA dataset. Based
39
on these observations and the previous observation about the hardness of the three da-
tasets (AM > BR > UA) we conclude, Jiang similarity is the most suitable word-level
semantic similarity measure.
Performance variation across sentence-level methods. From Table 2 it can be observed
that for the UA dataset, among the five word-level measures, Lin similarity obtained
the highest F1 score with Optimal and QAP pairing (i.e. 0.520 for Optimal and 0.525
for QAP pairing). However, in the case of Greedy pairing sentence-level method, both
Resnik and Jiang measures obtained a higher F1 score than Lin similarity. Similarly, for
the BR dataset, both Lin similarity and Jiang similarity obtained the highest F1 score
with QAP pairing. However, the F1 scores obtained by Jiang similarity with Optimal
and Greedy pairing is higher than that of the Lin similarity. These changes in the best
performing similarity measures due to the change in sentence-level methods, highlights
the significance of sentence-level methods. Hence, we conclude that adequate attention
should be given to the choice of the sentence-level methods. Another key observation
regarding the sentence-level methods is that, for each dataset, the highest F1 score ob-
tained by a word-level measure involved QAP pairing. This indicates that QAP pairing
is the most suitable sentence-level aggregation method than Optimal and Greedy pair-
ing methods.
5 Related Work
A plethora of process model matching techniques have been developed which can be
broadly divided into two types, syntactic and semantic [5]. Syntactic techniques merely
rely on the similarity or distance between the labels without taking into consideration
the meaning of the words. In contrast, semantic techniques rely on the semantics of
words for computing similarity.
A recent survey of PMM has identified a set of semantic matching techniques that
are used in literature [5]. A summary of these techniques is presented in Table 3. In the
table, Wu, Leacock, Jiang represents Wu & Palmer, Leacock & Chodorow and Jiang
& Conrath word-level semantic similarity measures. The โ+โ sign in the table indicates
that the technique uses the respective technique, whereas the โโโ sign indicates that the
technique is not used in the paper. There are occasions in which the use of synonyms is
implicit, these are marked as โ+/โ.
Table 3. Semantic matching techniques using for computing similarities between activity pairs
Author et al. Synonyms Lesk Wu Leacock Resnik Jiang Lin
Sebu et al. [25] + + - - - - +
Sonntag et al. [26] + - - - - - +
Sebu et al. [28] + + - - - - +
Makni et al. [29] + - - - - - -
Fengel [30] - - - - - - -
40
Pittke et al. [31] + - - - - - +
Klinkmuller et al. [32] - - - - - - +
Klinkmuller et al. [33] + - + - - - +
Klinkmuller et al. [34] + - - - - - +
Jin et al.[35] + - - - - - -
Caygolu et al. [36] - - + - - - -
Belhoul et al. [37] - - + - - - -
Niemann et al.[38] + - - - - - -
Leopold et al. [39] + - - - - - +
Humm et al. [40] +/- - - - - - -
Dijkman et al. [41] +/- - - - - - -
Dumas et al. [42] + - - - - - -
Dongen et al. [43] +/- - - - - - -
Agnes et al. [44] + - + - - - -
Ehrig et al. [45] + - + - - - -
Corrales et al. [46] + - - - - - -
Schoknecht et al. [8] - - - - - + -
From the table it can be observed that most of the studies propose to use synonyms
for semantic similarity. However, these studies do not explicitly present the measures
used for computing similarity. Also, it can be seen from the table that, Lesk, Wu and
Lin are the other similarity measures used in literature. Furthermore, it can be observed
that Leacock, Resnik and Jiang measures have never been used for identifying corre-
sponding activities between a pair of process models. Additionally, only word-level
semantic similarity measures are considered for computing similarities between activity
pairs and these word-level similarity measures have not been extended to compute sim-
ilarity at activity-level.
6 Conclusion
Several semantic Process Model Matching (PMM) techniques have been proposed,
however these techniques merely focus on the word-to-word semantic similarity, with-
out due consideration to aggregation of word-level similarity to sentence-level (or ac-
tivity-level) similarity. Furthermore, the existing studies have only used three semantic
similarity measures and ignored the other semantic similarity techniques that have
shown promising results for various text processing tasks. To that end, in this paper, we
have used five word-level sematic similarity measures and three sentence-level aggre-
gation techniques to experimentally evaluate the effectiveness of all the 15 combina-
tions in the context of PMM. For the experiments we have used well established da-
tasets from PMMCโ15. The results reveal the following: a) the hardness of the three
41
datasets are different, with AM dataset being the hardest, BR dataset being the moder-
ate, and UA dataset being the easiest, b) Jiang similarity, is the most suitable matching
technique, and c) QAP Pairing is the most effective sentence-level measure. In the fu-
ture, we plan to compare the performance of these semantic measures with all the ex-
isting matching techniques.
References
1. Kuss, E., Leopold, H., et al: Probabilistic Evaluation of Process Model Matching
Techniques. In: Proceedings of the 35th International Conference on Conceptual
Modeling, pp. 279-292, Gifu, Japan, (2016)
2. Cayoglu, U., Dijkman, R., Dumas, M. et al.: Report: The Process Model Matching
Contest 2013. In: Proceedings of the Business Process Management Workshops,
Springer LNBIP, pp. 442-463, Beijing, China, (2013).
3. Meilicke, C., Leopold, H., et al: Overcoming Individual Process Model Matcher
Weaknesses Using Ensemble Matching. DSS, 100 (1), pp. 15-26, (2017).
4. Kuss, E., Leopold, H., et al: Ranking-based evaluation of process model matching.
In: Proceedings of the International Conference on Cooperative Information Sys-
tems, pp. 298-305, Rhodes, Greece, (2017).
5. Jabeen, F., Leopold, H., Reijers, H. A.: How to make process model matching work
better? An analysis of current similarity measures. In: 20th International Confer-
ence on BIS, pp. 181-193. Springer LNBIP, Poznan, Poland, (2017).
6. Leacock, C., Chodorow, M.: Combining local context and WordNet similarity for
word sense identification. In: WordNet: An electronic lexical database, MIT Press,
(1998).
7. Resnik, P.: Using information content to evaluate semantic similarity in a taxon-
omy. In: Proceedings of the 14th International Joint Conference on Artificial Intel-
ligence, pp. 448-453. Quebec, Canada, (1995).
8. Schoknecht, A., Thaler, T., Fettke, P., Obserweis, A., Laue, R.: Similarity of Busi-
ness Process models โ A State-of-the-Art Analysis. ACM Computing Survey,
50(4), pp. 1-33, (2017).
9. Lin, D.: An information-theoretic definition of similarity. In: Proceedings of the
15th International Conference on Machine Learning, pp. 296-304, (1998).
10. Wu, Z., Palmer, M.: Verbs semantics and lexical selection. In: Proceedings of the
32nd Annual Meeting on ACL, pp. 133-138, Las Cruces, New Maxico, (1994).
11. Budanitsky, A., Hirst, G.: Evaluating WordNet-based measures of lexical semantic
relatedness. Computational Linguistics, 32(1), pp. 13-47, (2006).
12. Navigli, R.: Word sense disambiguation: A survey. ACM Computing Surveys,
41(2), pp. 1-12, (2009).
13. Androutsopoulos, I., Malakasiotis, P.: A survey of paraphrasing and textual entail-
ment methods. Journal of AI Research, 38 (1), pp. 135-187, (2010).
14. Gabrilovich, E., Markovitch, S.: Computing semantic relatedness using Wikipe-
dia-based explicit semantic analysis. In: Proceedings of the 20th International Joint
Conference on Artificial Intelligence, pp. 1606-1611, Hyderabad, (2007).
42
15. Lord, W.P., Stevens, R.D, Brass, A., Goble, C.A.: Investigating semantic similarity
measures across the Gene Ontology: the relationship between sequence and anno-
tation. Bioinformatics, 19(10), pp.1275-1283, (2003).
16. Farhadi, A., Hejrati, M., et al.: Every picture tells a story: Generating sentences
from images. In: Proceedings of the European Conference on Computer Vision,
Springer LNCS, pp. 15-29, Crete, Berlin, (2010).
17. Androutsopoulos, I., Malakasiotis, P.: A survey of paraphrasing and textual entail-
ment methods. Journal of AI Research, 38 (1), pp. 135-187, (2010).
18. Liu, B.: Sentiment analysis and opinion mining. Morgan & Claypool, (2012).
19. McCallum, A., Xuerui, W., Corrada-Emmanuel, A.: Topic and role discovery in
social networks with experiments on enron and academic email. Journal of AI Re-
search, 30 (1), pp. 249-272, (2007).
20. Chang, J., Garber, J.B., Gerrish, S., Wang, C., Blei, D.M.: Reading tea leaves: How
humans interpret topic models. In: Proceedings of the 22nd International Confer-
ence on NIPS, pp. 288-296, BC, Canada, (2009).
21. Corley, C., Mihalcea, R.: Measuring the semantic similarity of texts. In: Proceed-
ings of the ACL Workshop on empirical modeling of semantic equivalence and
entailment, pp. 13-18, Michigan, USA, (2005).
22. Rus, V., Lintean, M.: A comparison of greedy and optimal assessment of natural
language student input using word-to-word similarity metrics. In: Proceedings of
the Seventh Workshop on Building Educational Applications Using NLP, pp. 157-
162, Montreal, Canada, (2012).
23. Lintean, M., Rus, V.: An Optimal Quadratic Approach to Monolingual Paraphrase
Alignment. In: Proceedings of the 20th Nordic Conference of Computational Lin-
guistics, pp. 127-134, Vilnius, Lithuania, (2015).
24. Antunes, G., Bakhshandeh, M., Borbinha, J., Cardoso, J., Dadashnia, S., Di Fran-
cescomarino, C. & Hake, P. (2015). The process model matching contest 2015. GI-
Edition/Proceedings: Lecture notes in informatics, 248, 127-155.
25. Sebu, M. L., Ciocรขrlie, H.: Similarity of business process models in a modular
design. In: Proceedings of the IEEE 11th Symposium on Applied Computational
Intelligence and Informatics, pp. 31-36, Timisoara, Romania, (2016).
26. Sonntag, A., Hake, P., Fettke, P., Loos, P.: An approach for semantic business pro-
cess model matching using supervised learning. In: Proceedings of the European
Conference on Information System, pp. 1-10, Istanbul Turkey, (2016).
27. Sebu, M. L., Ciocarlie, H.: Merging business processes for a common workflow in
an organizational collaborative scenario. In Proceedings of the 19th International
Conference on STCC, pp. 134-139, Cheile Gradistei, Romania, (2015).
28. Makni, L., Haddar, N.Z., Ben-Abdallah, H.: Business process model matching: An
approach based on semantics and structure. In Proceedings of the 12th International
Conference on e-Business and Telecommunications, pp. 64-71, (2015).
29. Fengel, J.: Semantic technologies for aligning heterogeneous business process
models. Business Process Management Journal, 20(4), pp. 549-570, (2014).
30. Pittke, F., Leopold, H., Mendling, J., Tamm, G.: Enabling reuse of process models
through the detection of similar process parts. In: Proceedings of the BPM Work-
shops, Springer LNBIP, pp. 586-597, Tallinn, Estonia, (2012).
43
31. Klinkmรผller, C., Leopold, H., et al.: Listen to me: Improving process model match-
ing through user feedback. In: Proceedings of the International Conference on
BPM, Springer LNCS, pp. 84-100, Haifa, Israel, (2014).
32. Klinkmรผller, C., Weber, I., Mendling, J. et al.: Increasing recall of process model
matching by improved activity label matching. In: Proceedings of the International
Conference on BPM, Springer LNCS, pp. 211-218, Beijing, China (2013).
33. Jin, T., Wang, J., La Rosa, M., Ter Hofstede, A., Wen, L. Efficient querying of
large process model repositories. Computers in Industry, 64(1), pp.41-49, (2013).
34. Cayoglu, U., Oberweis, A., Schoknecht, A., Ullrich, M.: Triple-S: a matching ap-
proach for Petri nets on syntactic, semantic and structural level. In: Proceedings of
PMMCโ13, co-located with BPM, Beijing, China (2013).
35. Belhoul, Y., Haddad, M., et al.: String comparators based algorithms for process
model matchmaking. In: Proceedings of the International Conference on Services
Computing (SCC), pp. 649-656, Honolulu, USA, (2012).
36. Niemann, M., Siebenhaar, M., et al.: Comparison and retrieval of process models
using related cluster pairs. Computers in Industry, 63(2), pp. 168-180, (2012).
37. Leopold, H., Niepert, M., Weidlich, et al: Probabilistic optimization of semantic
process model matching. In: Proceedings of the International Conference on BPM,
pp. 319-334, Tallinn, Estonia, (2012).
38. Humm, B. G., Fengel, J.: Semantics-based business process model similarity. In:
Proceedings of the International Conference on Business Information Systems, pp.
36-47, Vilnius, Lithuania, (2012).
39. Dijkman, R., Dumas, M., et al.: Similarity of business process models: Metrics and
evaluation. Information Systems, 36(2), pp. 498-516. (2011).
40. Dumas, M., Garcia-Banuelos, L., Dijkman, R.M. Similarity search of business pro-
cess models. IEEE Data Eng. Bull., 32(3), pp. 23-28, (2009).
41. Van Dongen, B., Dijkman, R. et al.: Measuring similarity between business pro-
cess models. In: Proceedings of the CAiSE, pp. 405-419, Valencia, Spain, (2013).
42. Koschmider, A., Oberweis, A.: How to detect semantic business process model
variants? In: Proceedings of the ACM symposium on Applied computing, pp.
1263-1264, Seoul, Korea, (2007).
43. Ehrig, M., Koschmider, A. et al.: Measuring similarity between semantic business
process models. In: Proceedings of the 4th Asia-Pacific conference on Conceptual
modelling, Springer LNCS, pp. 71-80, Ballarat, Australia, (2007).
44. Corrales, J., Grigori, D. et al.: BPEL processes matchmaking for service discovery.
In: Proceedings of the CoopIS, pp. 237-254, Montpellier, France, (2006).
44