Documente Academic
Documente Profesional
Documente Cultură
a r t i c l e in f o abstract
Article history: A designer's information need can have a lot of finer points not expressible completely
Received 11 November 2013 by textual attributes or global aggregated image features. The purpose of the study is to
Received in revised form develop a cognitive environment for content-based engineering drawing retrieval,
24 December 2014
allowing a user to concretize queries, to implement strategies that are most effective
Accepted 26 December 2014
for the current search task. We propose meaningful customizable graphic search patterns,
Available online 5 January 2015
various comparison modes and visualization of matched elements. The presented
experiments show that the approach yields good results in the sense of relevance and
is comfortable in terms of controllability. The possibility to customize search patterns can
Keywords:
Technical drawing result in improving retrieval precision by 10–15%. In practice, it can significantly increase
Content-based retrieval the degree of automation of the design-reuse process.
Matching graphs & 2015 Elsevier Ltd. All rights reserved.
Search strategy
http://dx.doi.org/10.1016/j.jvlc.2014.12.010
1045-926X/& 2015 Elsevier Ltd. All rights reserved.
D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146 135
it in other cases, to modify the search prescription without on the base of computing its spectrum. To speed up
redrawing the sample image, to set exact requirements to searching, descriptors are indexed on the base of the NB-
similarity of individual components, etc. Tree. The retrieval process is divided into two steps: (1)
Search engines' users can be divided into three classes, selecting drawings topologically similar to the query and
depending on how clear they understand their search (2) computing geometry similarity between each candi-
need [3]: (1) a browser has no clear end-goal, performs a date drawing and the query sketch.
series of unrelated searches; (2) a surfer does exploratory The method described in [10] is based on the Hough
searches in the beginning, but in the process of working transform, which is applied for generation of a feature
his clarity of what he wants from the system is gradually vector. Not the whole cumulative matrix is used, which is
increased; (3) a searcher has a clear intent, performs well built when calculating the Hough transform, but only the
related searches leading to an end-result. The same user in angular information. Comparison of images is performed
different search sessions may relate to different classes, by calculating a distance between their feature vectors
and his search strategies vary considerably. Therefore, a according to a very simple formula.
good retrieval system should provide an opportunity for The approach proposed in [11] represents drawings as
each class of users to build effective strategies, including feature vectors of pixel blocks and calculates their simi-
individual for each person. larity by the linear weighted cosine similarity method.
The purpose of the study is to develop a retrieval In the method described in [12] an elementary descrip-
environment allowing a user to concretize queries, to tor of a drawing image is a so-called local neighborhood
implement various strategies that are most effective for structure (for simplicity, also called as patch), which is
the current search task. It is necessary to expand the obtained by grouping lines/arcs around some reference
existing scenario of the drawing retrieval dialog. line. Matching is based on identification of dense regions
The paper is organized as follows. In Section 2 related in a transformation parameter space (object center, scale,
works on content-based technical drawing retrieval are ana- rotation) using the mean shift method.
lyzed. In Section 3 we describe our approach for automatically The system CADFind [13] indexes not only geometry of
describing and matching drawing images, propose the retrie- the part in a drawing, but also text information associated
val environment's elements that are aimed to improve the with it, including material, manufacturing process. Infor-
retrieval flexibility and relevance. In Section 4 the retrieval mation extracted from a drawing is represented in the
system's prototype is described, experiments on evaluation of form of a so-called Group Technology (GT) code, a string of
the proposed retrieval environment are provided. In Section 5 textual–numeric symbols. Matching drawings is carried
we discuss obtained results. Finally, in Section 6 the conclu- out by fuzzy comparison of corresponding GT codes.
sion is made and directions for further work are listed. There are also approaches that use a structural descrip-
tion of a drawing's graphic object, but matching drawing
2. Related work images has a mechanical bias; there is no meaningful
analysis of components and relations between them. In the
There are many approaches to the content-based tech- paper [14] a method for searching technical drawings on
nical drawing retrieval. Extensive surveys are even the base of matching attributed graphs describing draw-
devoted to them [4,5]. ings is proposed. Graph nodes correspond to primitives
In the area of graphical retrieval of drawings and in (lines and curves), and edges describe spatial relations
pattern recognition in general, the approaches prevail in between these primitives. To solve the graph matching
which a global description is given to a graphic object, and problem, the mean field theory is used.
comparison of graphic objects consists in applying some In the paper [15] two alternative methods for searching
algebraic distance measure to their aggregated descriptions. raster technical drawings in a patent database are considered:
In the experimental system ShapeLab [6] two drawing by measuring the similarity of graphs and by comparing
matching methods are implemented. In the first method, histograms. Both methods involve transformation of the line
drawings are represented as spherical harmonics; a fast set of a drawing into directed graph of 6-nearest (by Euclidean
transformation is applied for their comparing. In the second distance between line centers) neighbors, edges of which are
method, a drawing is represented as a 2D shape histogram, a loaded by invariant attributes: relative angle and relative
distribution of distances between pairs of randomly selected position of two lines. These attributes are also used when
points; for measuring similarity between two such histo- two-dimensional histograms are created. Matching graphs is
grams the Minkowsky distance is used. carried out with the use of a variant of the Hausdorff distance.
In the work [7] an approach for retrieving vector Comparison of histograms is performed by measuring the
drawings by a sketch drawn by a hand on a tablet device Bhattacharya correlation.
is presented; earlier versions of this approach are consid- The system that is proposed in [16] represents drawing
ered in [8,9]. The content of drawings is decomposed into images in the form of hierarchical topology graphs and
topology (spatial arrangement of polygons in a drawing) performs their matching on the base of nested assignment
and geometry (shape of these polygons). Topological algorithm and the EMD distance.
information is organized in the form of a graph, in which The purpose of the work [17] was to carry out matching
nodes represent the drawing's polygons, and edges corre- parts' contours in such a way that the matching result
spond to relationships between them of three types: is adequate to engineering interpretation: determining
inclusion; adjacency; and spatial proximity. The topology overall structural similarity with tolerance to local bending
graph is then converted into multidimensional descriptors and stretching. The two-level representation of shapes is
136 D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146
used: a set of feature points with their local geometry, usage of standard mechanical engineering elements
which represent high-level structure of a contour, and and qualitative characteristics when describing graphic
groups of elements between feature points, which repre- objects;
sent low-level geometric features of the contour. Connec- automatic extraction of 3D constructive objects from a
tion between the levels is organized by means of a lookup drawing in order to describe it by them;
table. Feature points are determined among endpoints of a representation of a graphic object at various levels of
contour's elements. Contour matching consists in finding abstraction;
the optimal correspondence of feature points and groups tolerance to affine as well as topological transforma-
of elements between them. tions of compared objects; and
The querying process in the existing approaches is fixation and visualization of similarities and differences
limited by specifying a sample image; searching is carried between graphic objects.
out at the level of abstraction that was predefined by
developers and might not correspond to a user's need; at The base model of representation of a graphic object is the
that the user has no possibility to determine the appro- fuzzy spatial loaded graph [18,19] (FSLG) G¼ (V, AV, E, AE, AO),
priate set of search features. This eventually may adversely where V is the set of nodes that describe components of the
affect search results. object using the parameters AV, E is the set of edges that
In our view, the possibility to use careless sketches represent spatial relations between the components using the
as queries [6,7] is good only for simple parts. For creating an parameters AE, AO is global parameters of the object. One of
image of a complex part a designer more likely will resort to a the components' parameter is similarity with typical config-
CAD system in which will create a desired sketch. Further- urations; this improves the representation's meaningfulness,
more, searching is often performed not at the beginning, but allows performing indirect, faster matching of components.
during the process of developing a project, when there is The parameters AV, AE, AO are represented as linguistic
already a rough drawing of a designed part. variables [20]; this also allows increasing matching speed,
In the design field a search by an image-fragment is quite enhancing analysis and matching abilities, makes the repre-
important. This possibility is not provided at all in the sentation more resistant to deformations.
approaches [6,10,11,13,17]. In the works [7,16] it is restricted Four levels of the representation of graphic information
by an image type – namely an image-fragment should be a are given: (1) graph of singular points and segments, (2)
composition of contours. It is doubtful that the considered graph of contours, (3) graph of 3D objects, (4) graphs of
approaches allow retrieving parts that have a given mechan- heterogeneous components. Unlike aggregated descriptors
ical feature (such as groove, keyseat, chamfer with a certain used in many related works, the listed representations are
angle, fillet of a specified radius, etc.). At the same time, this understandable by humans. The stratification into layers
search kind is demanded in developing technological process allows fixing various, low- and high-level aspects of the
of manufacturing a part, since exactly mechanical features same drawing image.
complicate the manufacturing process. Like in most related works, the drawing indexing stage
In our view, for a user it is easier to interact with such a is provided in our approach, in which search patterns of
search system that provides him with information about drawings are created and entered into a searchable data-
what, from its point of view, is the same and what is base by which searching is further performed.
different. Establishing correspondences between image Details of our models, methods and algorithms are
elements is implemented only in the approaches [16,17]. presented in [21–24].
We share the opinion of [5] that applied to most of the
developed approaches high recognition quality is reported,
3.1. Query drawing types
whereas they were evaluated at a quite limited amount of
drawings; their capabilities must be assessed in large-scale
An existing detailed drawing, a CAD draft or a sketch
experiments.
may serve as a sample image. A drawing format may be
In this paper we present and evaluate a new approach
both vector and raster. A raster drawing can be automati-
to the drawing retrieval problem, which is based on graph
cally converted to a vector form by means of an automatic
matching. For the purpose of improving retrieval flexibility
vectorizer; our technology is presented in [25].
and relevance, the dialog between a user and a search
Due to the variety of the representation levels (see
system is carried out by means of meaningful customiz-
Section 3.2), a designer may build a query on the base of a
able search patterns of different abstraction level, with
sample image of any content: a projection view, an
help of visualization of matched elements, which is pro-
external contour, a fragment of a projection, a standard
vided by our graph matching method.
mechanical feature (see Fig. 1).
3. Cognitive retrieval environment 3.2. Search pattern kinds and their application areas
The key conceptions underlying our approach to the In the retrieval system drawing images are represented
technical drawing retrieval problem are as follows: as search patterns of the four types:
consideration of a graphic object as a multi-level system (1) Graph of singular points and segments describes a draw-
like a linguistic one; ing image at the level of chains of primitives. Graph nodes
D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146 137
contour shape: square, rectangle, rectangle with rounded From a user's point of view, the presence of certain
corners, rectangle with truncated corners (etc. according elements in target drawings is a decisive factor, the
to the library of typical contours); presence of other is preferable but is not necessary, while
D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146 139
some elements should be absent at all. For this reason, the allow a user to flexibly (depending on a task) customize a
possibility to specify obligatory and optional components search.
as well as elements that should not be in result drawings is
provided in all the search pattern types. If at least one of 4. Experiments
obligatory components was not detected in a drawing or it
was found that the drawing contains at least one of 4.1. Implementation – the system GrSearch
prohibited items, then the drawing is considered as not
corresponding to the query. Our approach is implemented in the system for gra-
In the matching process the system considers various phical searching vector mechanical drawings, which is
variants of combination of segments through insignificant called GrSearch. Fig. 6 shows its search dialog window,
angles. The most optimal variant of combination is the one in which a search pattern that will be used in the search is
that does not lose details and leads to the best final result selected and adjusted.
of the matching according to the recursive preliminary Fig. 7 provides examples of the visualization of
analysis [24]. The retrieval system automatically assigns matched drawing elements by the system.
significant angles (depending on angle values and lengths
of connected segments). The user also has the possibility
4.2. Refinement of a search
to manually change the set of significant/insignificant
angles. The matching algorithm's ability to combine seg-
We evaluated how much search quality improves when
ments through insignificant angles allows identifying
obvious concretization of a query is made. Experiments
similarities between objects that have considerable local
were carried out on a test database containing 4000
deviations from each other.
drawings, in which almost 7000 projection views were
The evaluations of the impact of the described para-
presented. Four search types were carried out:
meters of search patterns to the retrieval performance and
quality are depicted in Table 1.
search by an external contour without query concreti-
Working with search patterns' parameters is carried
zation,
out mostly by mouse: hypergraphics, lists of qualitative
search by an external contour with query refinement,
values to choose from. So, the system's interface is intui-
search by a graph of contours with default para-
tive and easy to learn.
meters, and
search by a customized graph of contours.
3.4. Developing and implementing a search strategy Basing experiments on such search patterns as external
contours and graphs of contours is caused by the fact that
Consider the work process of a user in the retrieval they are the most frequently used. The study of each
system within a particular search session. Generally, before search type included execution of 50 queries; sample
the search, the user creates an initial query, based on his/ images used in each search type were the same; CAD
her experience and our recommendations on using the drafts were used in the queries. The retrieval system was
system, and in the process of implementing his/her strat- run on the computer Pentium E6800 3.33 GHz, 2 GB DDR3.
egy refines the query, including a return to early stages, Query refinement was directed towards both tighten-
until a satisfactory result is obtained. ing and weakening requirements to search results.
In Fig. 5 the scenario of the dialog on refining a query is Fig. 8 shows precision/recall graphs for performed
considered. The query refinement allows improving search searches, which were built according to the 11-Point
relevance and/or speed, narrowing or expanding a search. Interpolated Average Precision method [26]. They show
As can be seen from the diagram, the proposed envir- that due to the refinement of queries the average retrieval
onment provides a user with many ways (strategies) of precision increased, especially at the beginning – by
solving a search task. The developed tools are aimed to 10–15%.
Table 1
The impact of some parameters to searching.
Parameter Change of the Search The probability of missing The probability of obtaining
parameter speed relevant documents irrelevant documents
In some cases there were no significant changes in The experiments also show that the system is well tuned
search results, while in many situations changes were very and provides quite adequate results by non-refined queries.
noticeable and positive.
The average number of additional directives that were
made by a user in the query refinement process is 7.5 for 4.3. User study
the searches by external contours and 5.2 for the searches
by graphs of contours; one directive is given by two mouse In order to evaluate the system from the user perspec-
clicks. tive, we involve employees of design bureaus of a manu-
The degree of involvement of query refinement possi- facturing enterprise. A total of 5 designers were involved.
bilities in the searches can be seen in Table 2. The following tasks were assigned to the designers:
In 36% of the searches by external contours, irrelevant
search results were reduced by more than 2 times after the (1) Evaluate results of a search conducted by a query with
query refinement. In 60% of the searches by graphs of default parameters, on the scale of 0 to 10.
contours, irrelevant search results were reduced by more (2) Evaluate results of a search, conducted with the query
than 10 times after the query concretization. customization, on the scale of 0 to 10.
The average recall of the searches by external contours (3) Indicate, what features were used in the query refine-
without query refinement is 67.8%, and with the customiza- ment process.
tion is 74.9%. The average recall of the searches by graphs of (4) Indicate, whether the search results are necessary for
contours is 71.8% (without the query concretization) and carrying out the design/production engineering.
75.4% (with the query refinement). A relevant document was (5) Indicate, whether the search results are necessary
considered as a positive detection if it was present on the for studying the experience of the design/production
first result page that holds up to 75 drawings. engineering.
The average speed of the search by an external contour
without refinement is 44 s, and with the query refinement In one experiment the initial search by an external
is 36 s. The average speed of the search by a graph of contour of a part gave a result set of 211 drawings, which
contours without the query refinement is 1 min 32 s, and meet the expectations of a specialist (10 points). After
with the query refinement is 1 min 18 s. refinement of the query (the length and angle are exactly
Figs. 9 and 10 show results of some searches that were the same as that of the sample), the number of results
conducted in the experiment. As we can see, the results reduced to 6. On the one hand, these results correspond
obtained by concretized queries better correspond to the to the declared parameters. On the other hand, some
sample images. very similar drawings having other dimensions of the
D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146 141
mentioned parameters are disappeared from the results. results, the users’ opinion was unequivocally positive in
As a result, the search with the query refinement was most cases.
evaluated by only 6 points. However, it was noted that, in The users made the following proposals for further
general, specification of parameters positively affects pre- developing the system:
cision and speed of the search.
In other experiment the initial search by a graph of
contours (of a projection view of a part) gave a result set of To add possibility of searching a drawing by its verbal
9 drawings, which was assessed by a specialist by 7 points. description in order to simplify the querying process.
After refining the query (the presence of a contour is To implement means of reducing a search space. For
obligatory, a contour's shape is the same) the number of example, knowing that there is one or another element
results reduced to 4, and the retrieved results were (a hole, a groove, etc.) in a drawing, it will be possible to
estimated by 10 points. limit the retrieval by the presence of such an element.
Among the query refinement features only the possi- Creating a separate searchable database by each bureau
bility of excluding a contour from consideration was asse- and using the system within a bureau.
ssed as unclaimed. Regarding the usefulness of retrieved Increasing the search speed.
142 D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146
Fig. 8. Precision/recall for the searches with and without query refinement.
The system can find similarities between images, this allows a user to implement effective individual strategies
regardless that they differ significantly on the base of local of searching graphical design-engineering information.
stretching and/or bending. In this respect we have moved We identify the following ways of improving the
a little further than the work [17], in which the need for retrieval environment:
such a search mode was raised for the first time. In our
approach it is possible to significantly narrow or expand automatic synthesis of graphic search patterns by a
the boundaries of permissible differences between parts. verbal description of a drawing;
We believe that the possibilities of adjusting the pre- automatic generation of recommendations on search-
cision of searching, performing the search at different ing on the base of fuzzy rules built on graphic patterns
abstraction levels help a user to be sure that none of of common classes of mechanical parts;
relevant documents will be overlooked. automatic tuning of search patterns on the base of
relevance feedback, component correspondence marks
and machine learning methods; and
6. Conclusion increasing the search speed.
The proposed environment for graphical searching techni- We hope that in the near future complex retrieval
cal drawings has the following characteristics: various levels solutions like ours will enter designers' arsenal of usual
of meaningful representation of graphic objects, the possibility means. For example, it has already happened with the
to concretize queries in qualitative concepts, various compar- content-based 3D model retrieval system IMShape [27]
ison modes, and visualization of matched elements. Together, which was incorporated in the regulations of a Russian
144 D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146
Table 2
Refinement possibilities and their involvement in searches.
enterprise as a mandatory preliminary design phase (see workplace, J. Am. Soc. Inf. Sci. Technol. 60 (3) (2009) 443–454,
[28]). The result of introduction of such solutions is the http://dx.doi.org/10.1002/asi.v60:3.
[2] D. Tunkelang, Reconsidering relevance and embracing interaction,
significant increase in the degree of automation of the
Bull. Am. Soc. Inf. Sci. Technol. 36 (1) (2009) 20–23, http://dx.doi.
design-reuse process, reduction of the labor intensity of org/10.1002/bult.2009.1720360107.
searching the documents containing graphic elements, [3] R. Datta, D. Joshi, J. Li, J.Z. Wang, Image retrieval: ideas, influences,
and reduction of new product design time and costs. and trends of the new age, ACM Comput. Surv. 40 (2) ;40(2008),
http://dx.doi.org/10.1145/1348246.1348248. (Article 5, 60 pp.).
[4] N. Weber, A. Henrich, Retrieval of technical drawings in DXF format –
concepts and problems, in: LWA 2007 Workshop Proceedings, 2007,
Acknowledgments pp. 213–220.
[5] N. Bhatti, A. Hanbury, Image search in patents: a review, Int. J. Doc.
Anal. Recognit. 16 (4) (2013) 309–329.
This work was partially supported by the state assign- [6] J. Pu, K. Ramani, On visual similarity based 2D drawing retrieval,
ment of the Ministry of Education and Science of the Comput.-Aided Des. 38 (3) (2006) 249–259, http://dx.doi.org/
10.1016/j.cad.2005.10.009.
Russian Federation on the subject of “Methodology and a [7] P. Sousa, M.J. Fonseca, Sketch-based retrieval of drawings using
system of retrieving, semantic referencing and extracting spatial proximity, J. Vis. Lang. Comput. 21 (2) (2010) 69–80, http:
knowledge from Russian- and English-language text and //dx.doi.org/10.1016/j.jvlc.2009.12.001.
[8] M.J. Fonseca, J.A. Jorge, Towards content-based retrieval of technical
graphic documents (in various information resources)”. drawings through high-dimensional indexing, Comput. Graph. 27 (1)
(2003) 61–69, http://dx.doi.org/10.1016/S0097-8493(02)00244-3.
References [9] M.J. Fonseca, A. Ferreira, J.F. Jorge, Content-based retrieval of technical
drawings, Int. J. Comput. Appl. Technol. 23 (2) (2005) 86–100.
[10] P. Fränti, A. Mednonogov, V. Kyrki, H. Kälviäinen, Content-based
[1] S. Allard, K.J. Levine, C. Tenopir, Design engineers and technical matching of line-drawing images using the Hough transform, Int. J.
professionals at work: observing information usage in the Doc. Anal. Recognit. 3 (3) (2000) 117–124.
146 D.R. Kasimov et al. / Journal of Visual Languages and Computing 28 (2015) 134–146
[11] L. Jiao, F. Huang, Z. Teng, An engineering drawings retrieval method [19] H. Aboulmagd, N. El-Gayar, H. Onsi, A new approach in content-
based on density feature and improved moment invariants, in: based image retrieval using fuzzy, Telecommun. Syst. 40 (1–2)
ISIP'09 Proceedings, 2009, pp. 352–355. (2009) 55–66, http://dx.doi.org/10.1007/s11235-008-9142-9.
[12] R. Liu, Y. Wang, T. Baba, D. Masumoto, Shape detection from line [20] L.A. Zadeh, The concept of a linguistic variable and its application to
drawings with local neighborhood structure, Pattern Recognit. 43 (5) approximate reasoning, Inf. Sci. 8 (3) (1975) 189–281.
(2010) 1907–1916, http://dx.doi.org/10.1016/j.patcog.2009.11.022. [21] A.V. Kuchuganov, D.R. Kasimov, Graphical search of technical draw-
[13] D. Love, J. Barton, Aspects of design retrieval performance using ings over data warehouses, Prikl. Inform. 2 (2012) 84–92. (in
automatic GT coding of 2D engineering drawings, in: IDDME 2004 Russian).
Proceedings, 2004, DOI: 10.1080/09537280500180873. [22] D.R. Kasimov, A.V. Kuchuganov, A.E. Lopatkin, System of graphical
[14] R. Liu, T. Baba, D. Masumoto, Attributed graph matching based search of technical drawings, Intellekt. Sist. Proizv. 1 (2012) 152–157.
(in Russian).
engineering drawings retrieval, in: DAS 2004 Proceedings, LNCS
[23] A.V. Kuchuganov, D.R. Kasimov, Multilevel cognitive analysis in
3163, 2004, pp. 378–388, DOI:10.1007/978-3-540-28640-0_36.
graphical retrieval of drawings, Pattern Recognit. Image Anal. 23
[15] B. Huet, N.J. Kern, G. Guarascio, B. Merialdo, Relational skeletons for
(4) (2013) 518–523.
retrieval in patent drawings, in: ICIP 2001 Proceedings, vol. 2, 2001,
[24] D.R. Kasimov, A.V. Kuchuganov, Initial level of perception in the task
pp. 737–740, DOI:10.1109/ICIP.2001.958599.
of graphical search of technical drawings, in: OGRW-8-2011 Pro-
[16] P. Wang, S. Jiang, An engineering drawing retrieval method based on
ceedings, 2011, pp. 103–106.
nested assignment algorithm, J. Comput. Inf. Syst. 9 (2) (2013) [25] A.V. Kuchuganov, P.P. Oskolkov, Automation of processing and
713–720. semantic coding of digital images, Vestn. Komp. Inf. Tekhnol. 1
[17] S. Hou, K. Ramani, Structure-oriented contour representation and (2013) 41–44. (in Russian).
matching for engineering shapes, Comput.-Aided Des. 40 (1) (2008) [26] C. Manning, R. Prabhakar, H. Schütze, Introduction to Information
94–108, http://dx.doi.org/10.1016/j.cad.2007.08.007. Retrieval, Cambridge University Press, New York, 2008(496 pp.).
[18] R. Krishnapuram, S. Medasani, S.-H. Jung, Y.-S. Choi, [27] D. Kernitsky, IMShape – pick your model, SAPR Graf. 9 (2009) 4–6.
R. Balasubramaniam, Content-based image retrieval based on a (in Russian).
fuzzy approach, IEEE Trans. Knowl. Data Eng. 16 (10) (2004) [28] E.B. Baglayev, Unification – it's easy, 〈http://labktpp.ru/news2.html〉,
1185–1199, http://dx.doi.org/10.1109/TKDE.2004.53. 2013 (in Russian).