CC BY
14
Cloud of Science. 2019. T. 6. № 2 http:/ / cloudofscience.ru
The graph context-oriented ontological methods
G. V. Kanygin, M. S. Poltinnikova
Federal Center of Theoretical and Applied Sociology of the Russian Academy of Sciences st. 7-ya Krasnoarmeyskaya, 25/14, Saint-Petersburg, Russia, 190005
e-mail: g.kanygin@gmail.com; maria.poltinnikova@gmail.com
Abstract. Earlier the authors proposed the graph context-oriented ontological (GCOO) methods that make the tools of object-oriented programming (OOP) accessible for practical use by non-specialists in computer science. Now the authors present the workings of a test program (ontoeditor) that implements GCOO methods. The article shows how the mathematical model of GCOO methods, incomprehensible to non-specialists, boils down to intelligible user interfaces. In addition, the paper shows the user actions, which make the unprepared person able to describe social processes using natural language and, at the same time, check such descriptions with the help of OOP tools. All demonstrations run as examples of conceptualization of practical situations arising in the process of communication between social actors. Keywords: GCOO methods; branching; explanation; explanation context; elucidation context; generalization of concept; terminological graph (T-graph); graph of contexts hierarchy.
1. Introduction
Today, information and computer technologies (ICTs) are the tools that allow IT specialists to bring about the information society [1, 2]. Methods of object-oriented programming (OOP) are important components of the modern ICTs [3]. Given the positive experience of the practical application of ICT in building the information society, we would like to make the OOP methods explicit and accessible for ordinary participants of social processes.
In addition, we would like to contribute to ICT functionality according to the principles of qualitative data analysis (QDA) packages1. This software had already proved its effectiveness in numerous sociological studies aimed at resolving problems of mutual understanding between informants and sociologists [4, 5].
To improve the methods of describing social processes by their participants, we have proposed graph context-oriented ontological methods (GCOO methods) [6, 7]. One of their advantages lies in working with graphs whose vertices are natural language denota-
1 Atlas.ti http://www.atlasti.com/, MAXQDA http://www.maxqda.com/, NVivo http://www.qsrinternational.com/products_nvivo.aspx, Ethnograph http://www.qualisresearch.com/
tions. This is comprehensible to the users and does not require a labor-intensive training in formal specification language.
Graphs are widely used in modern ICTs for knowledge visualization. However, in such cases, they do not perform specification of conceptualization [8]. For example, the developers of UML 2.52 offer diagrams as a tool for describing the relationships between objects, but the diagrams themselves are built on the basis of a formal language. The language proves to be the main tool for the users, through which they can describe social interactions by virtue of relationships between objects under visualization.
There is another reason why we believe that direct use of OOP methods is welcome for modelers engaged in conceptualizing the social world. The OOP instruments allow their followers to put under control the elaboration of mutually appropriated conceptualization when modelers work as social actors. At present, developers of specification languages are ready to apply them for conceptualization of both physical and social worlds [9]. However, when conceptualizing social world, it is important to bear in mind that: (1) the participants in social processes themselves construct social reality [10]; (2) sociologist tracks sociological definitions to the level of individuals [11]. Thus, for sociological conceptualization, it is important not only to represent the object in the analytical form, but also to link this form to the person who proposed it.
It seems misleading to think about such concepts of the social world as beliefs, desires, or intentions as if they were coming from the physical world. Also, the very idea of modeling such concepts on the basis of pre-developed general-purpose ontological specifications (e. g. UFO-A and UFO-B) appears questionable [12].
The knowledge specification language for sociological applications should provide a way to describe the individuals and communities who carry out conceptualization in the process of social communication. Social communication is complex and multifaceted. Hence, a specification language in the field of sociology needs to allow the social actors to express the complexity of the modern information society [1, 2].
In general, there are three main advantages of GCOO methods that facilitate their possible usage for social world description. Firstly, they allow users working with natural language notations in ordinary mode. Secondly, the modeler builds the relations of the basic functionality of OOP, that is, modularity, encapsulation, polymorphism, compilation, etc., directly by operations on the graphs and does not use any specification language. Thirdly, the graphical interface allows to reduce the qualification requirements of the user who builds the ontological relationships.
The purpose of this article is two-fold. On the one hand, we aim to show how and to what extent the original GCOO syntax is accessible to the end user when performing con-
2
2 UML2.5: OMG Unified Modeling Language TM (OMG UML) Version 2.5 http://www.omg.org/spec/UML/2.5/PDF.
ceptual tasks common in the field of knowledge management. On the other hand, we wish to demonstrate how this syntax allows an unprepared author of knowledge to use OOP methods to describe social world.
We took the following five steps towards these goals. First, we present a math model that serves as an initial description of the proposed graph syntax. Second, we introduce a notation allowing the user to express the model relationships with the help of natural-language notation. Third, we demonstrate how the notation relates to the interfaces of the program written by us and how these interfaces allow user operating with GCOO methods. Next, we explain how the relationships of visibility and polymorphism arise thanks to the methods proposed. We also draw parallels between algorithms for generating structures, incorporated in GCOO methods, and compilation process used in programming. Finally, we propose an original mechanism for the assimilation of concept structures, which is the technique of defining their types in GCOO methods.
Since our research aims to contribute to the field of social communication methods, we develop GCOO methods on examples that affect real interactions. Here we consider an example of building collective knowledge of the Sociological Institute of the Russian Academy of Sciences (SI RAS). Being limited by the size of the article, we touch only some aspects of building knowledge of SI RAS, namely, the description of actors, the modeling of competences of co-authors of knowledge and the reuse of conceptual relationships.
2. How to present GCOO methods to user
GCOO methods may be described at several levels: the mathematical model, implementation of the model on Free Pascal with end-user interfaces, and examples of use. In this article we present a step-by-step overview of these levels appeared when building collective knowledge of the SI RAS.
2.1. The math mode
In this section we present main definitions of the math model of the context-oriented ontology (for details, see [6, 7]). A combination of words that the author uses for naming an object is called a concept. Dictionary is an unordered set of concepts
T = {ti, t2,..., tq}.
The author of ontology has to indicate an object through a concepts pair of dictionary: the first concept for naming the object, the second concept for describing the context, in which the object is seen.
Let P = {(xy),(x2y2),..., (xsys)} be a set of ordered pairs, introduced by the author at some time, where s is a number of pairs and xkyk are elements of T.
A two-level directed tree E with vertices from the set P x P is a branching:
E: (x, y) ^ { (x^ ), (y2),...,(xsys) }, 5 < q where s is a number of user-defined pairs at the lower level of the tree.
elucidation
Figure 1. The branching E with vertices from P xP
By definition x is the term, y is the context, (x, y) is the head pair. For k = 1,..., 5 xk is the explanation, yk is the explanation context, and (xk, yk) is the ground pair. The set of ground pairs {(x^,yk)},k = 1,...,5, is called the elucidation, it may be empty.
For example, the branching
(Actual list of employers, SI RAS)^{(Nancy Smith, SI RAS);(John Lincoln, SI RAS);(Bill Harrison, SI RAS)};
contains the head pair (Actual list of employers, SI RAS) and three ground pairs: (Nancy Smith, SI RAS), (John Lincoln, SI RAS), and (Bill Harrison, SI RAS). In these pairs Actual list of employers is the term, SI RAS is the context, Nancy Smith, John Lincoln, and Bill Harrison are explanations and SI RAS is the explanation context. The elucidation is (Nancy Smith, SI RAS);(John Lincoln, SI RAS);(Bill Harrison, SI RAS).
If the author creates a concept and does not include it into a pair, this concept exists in the dictionary and may be absent from the set of pairs. Therefore suppose that the dictionary T and the set of all branchings B satisfy the following condition: Vt eT 3E e B: t e E.
Suppose P is a set of all pairs of concepts involved in the branchings of the set B, then the structure {P, B} is called a context-oriented thesaurus (CO thesaurus).
The CO thesaurus is a digraph, in which the vertices are the elements of the set P and the edges are the edges of the branchings of the set B. Such a structure corresponds to the classical definition of ontology [8]. Then the CO thesaurus with the algorithms for constructing its various subgraphs is called a CO ontology.
Let tx be a context and there exists a branching (tx,i2) ^{...}. Then i2 is called a contextual extension of . The corresponding notation is .
For example consider branching
(Gender, Nancy Smith)^{(Female, Common Knowledge)}.
Here Nancy Smith is the context and there exist branching
(Nancy Smith, SI RAS)^{(Social&Demographic Characteristics, Russian Federation)},
therefore SI RAS is a contextual extension of Nancy Smith:
Nancy Smiths SI RAS.
If the set {y, y2,..., yn} contains all contexts of {P, B}, then the graph of contexts hierarchy (CH-graph) is a graph with vertices yk and with edges defined successively by virtue of contextual extensions.
To build up the main CO ontology structure, namely the terminological graph (T-graph), we need to take a root vertex (x,y) e P and a branching
E : (x,y) ^{(x,y),(x2,y2),...,(x,У)}, as well as to define a rule of coupling of contexts. We denote this rule f (y).
To build a T-graph we take from ground pairs of branching E( x, y) contexts y,y2,...,y such that they satisfy the rule f (y). Let the first level of the graph f (y) contains к < s contexts of the branching E(x,y): f (y) = {yrl,...,yrlc}. In this way, we get two-level T-graph:
(^ y) x^ yrl),...,(xrk, yrk }.
Then we repeat the operation: for every vertex of the first level, we look for a corresponding branching in {P, B}. If such a branching exists, it enters into the second level of graph under the condition that its context satisfies the rule f (y). In such a way, we increase the levels of T-graph up to the vertex of last generation, which has no branching in {P, B}.
2.2. Building a conceptual model with the ontoeditor
For users, context-oriented ontological methods of knowledge creation are hidden in the interfaces of the ontoeditor program, called Diagogue3. Here we wish to explain (a) semantic and (b) instrumental perspectives of using the ontoeditor when creating knowledge.
3 The program is written by the authors in Free Pascal 3.0.4. under the shell of Lazarus 1.8.0.
(a) When we set the problem of describing SI RAS, we want to answer the question: 'what is SI RAS?' We look for an answer constructed as a semantic network that describes this sociological object by the help of concepts and their mutual links. Answering the question with the help of the ontoeditor tools, we invent more concepts that link to each other through branchings. Essentially, creating branchings is a process of primary coding, well known in sociological studies [13-15]. Primary coding executed with the help of any QDA program is a very time-consuming part of the study [13-15]. Experienced sociologists recommend to perform the primary coding in as many details as possible [13-15]. GCOO methods considerably simplify this semantically complicated process and help avoid redundant coding.
(b) Let us explain the main features of the software implementation, which make the math constructions comprehensible for the non-mathematicians. To answer the question of what is SI RAS, knowledge author has to create a thesaurus, i. e a set of branchings, each of them represented as a two- level graph. To construct a branching, the user must have the dictionary, i. e the set of concepts with which he can populate any branching. The program's window, in which a user can create and collect the dictionary, as well as view the concepts, is shown in Fig. 2.
Я FGOSF ?017 i <=■ 1 ®
Тезаурус !• Понятие Э Идентификация <5 +
л Mnemo >
С J SI RAS
о COLLECTIVELY CREATED KNOWLEDGE
• STRUCTURE OF SI RAS □
X SUBJECT FIELD
STAFF OF SI RAS
• INHABITANTS OF CITY
•j CITY
•i COUNTRY
• ! INHABITANT DATA
• RUSSIAN ACADEMY OF SC ENCES
• MINISTERY OF EDUCATION AND SCIENCE
•i FEDERAL AGENCY OF SCIENCE ORGANIZATIONS
4 m ! ►
Figure 2. The dictionary form
The program allows the user to construct branchings by filling a special form. Fig. 3 demonstrates how this form looks like for the branching that starts a process of answering to the question of what is SI RAS.
(SI RAS, collectively created knowledge)^{(structure of SI RAS, ...); (guidance documents, ...); (subject field, SI RAS);(dissertation council, ...); (staff of RAS; ...)}.
For the ontoeditor's user, each relation E is represented as a subgraph in the three-level graph located on the form with which the user interacts. The number of root vertices of this graph is equal to the polysemy of the concept x. Each root vertex is associated with y, its children are the explanations x,X,•••, X• Each child x has exactly one own child, which relays to the context of the explanation y . The notion x, to which the whole set of branchings relates, is shown above the graph. We call such a representation the branching interface of the concept x.
[0]
3 :,acLULipeHWii ^OrmwH
SI RAS
5ft //COLLECTIVELYCREATED KNOWLEDGE ► 3ft STRUCTURE OF SI RAS 0 ANY CONCEPT ► GUIDANCE DOCUMENTS 0 ANY CONCEPT ► 5ft SUBJECT FIELD 5ft SI RAS ► m DISSERTATIONAL COUNCIL 0 ANY CONCEPT ► 3fi STAFF OF SI RAS 0 ANY CONCEPT
Figure 3. The branching form
Figure 3 represents the following relations between denotations of the math model and their view for user. x is SI RAS. SI RAS is a monosemantic concept, i. e. x has in thesaurus a single context y that is COLLECTIVELY CREATED KNOWLEDGE. x is STRUCTURE OF SI RAS and its context y is ANY CONCEPT. x2 is GUIDANCE DOCUMENTS and its context y2 is ANY CONCEPT. x3 is SUBJECT FIELD and its context y3 is SI RAS. x4 is DISSERTATION COUNCIL and its context y4 is ANY CONCEPT. At last, the final explication pair of concepts has the following view: x5 is STAFF OF SI RAS and its context y5 is ANY CONCEPT.
Thus, the math model becomes available to the user due to two features of the interface. First, the nodes of the graph are not variables, but direct natural-language notations. Second, during practical conceptualization, human being populates the branching interfaces by using drag & drop operations.
As a source of concepts for dragging operations, the user can use any forms visible on the screen. Among them, there can be dictionary forms, forms for the branching interfaces and forms for presentation of the generated graphs (see figures 4-8).
3. Features of the description of the actors
Now we going to show how the tools of the ontoeditor Diagogue work in the typical cases of creating knowledge of SI RAS. Information related to the state institutions, e. g. SI RAS, contains data on their employees. It is assumed in the GCOO methods that the employees are full-fledged co-authors of the knowledge about SI RAS. Particularly, these authors delineate themselves by themselves using in this process practically any structures they prefer. Let us look at how the end-users would implement such self-description.
Specification of the actor in the program. The elementary GCOO methods descriptor, earlier called concept (hereafter the word notion is used as a synonym), in a programmatic sense, is an object of the basic TNotion type. In addition to 'concept objects', GCOO methods allow their users to employ 'actor objects' during conceptualization. The latter belongs to its own TActor type that is a descendant of the basic TNotion specification.
As compared with TNotion type, TActor contains a complementary attribute aimed to specify the actor personally. By means of GCOO methods, this attribute can be expressed structurally and can be thought of as a personal profile of a TActor 'object'. Additionally, both object types have their associated glyphs, or tags, that distinguish them visually. For example, in Figure 4 we can see three actors: Nancy Smith, John Lincoln, and Harry Johnson. Profiles of each actor lie below the nodes associated with the actors themselves.
Structural description of the actor. The user of the GCOO methods always makes an actor's profile as a structure or graph. This approach differs from the one implemented in the database management system, where the profile is a set of data fields, each of its data type (integer, string, etc.). By contrast, the structure of GCOO descriptions is such, that all actors acquire descriptions which make them unique compared to others. There are several ways for the user to construct the structural variety of profile structures, but here we can consider only one example in detail. We introduce the concept Actual list of employers in the context of SI RAS using formula (1). Next, we define the list of actors whose structurally different profiles are under development.
(Actual list of employers, SI RAS)^{(Nancy Smith, ...);(John Lincoln, ...);(Harry Johnson, ...)} (1)
In order to build up the necessary structures, we enhance the thesaurus under development by adding corresponding branchings (Appendix B: 'empty' pairs (nothing, noth-
ing) omitted). In addition, we fill in the dictionary with concepts needed to create the branchings (Appendix A: in lexicographical order).
Figure 4. T-graph with actor profiles
Using the thesaurus created, we can generate a terminological graph that includes structurally different profiles, for example, for actors Nancy Smith and John Lincoln. We can see the structural differences of the actors' profiles in Fig. 5-7 which give three layers of profiles describing (Nancy Smith, SI RAS) and (John Lincoln, SI RAS).
Figure 5. Actor profiles layer 1
Figure 6. Actor profiles layer 2
Figure 7. Actor profiles layer 3
Organization of structural descriptions by means of contexts. The applied knowledge of SI RAS exists thanks to many social actors that contribute to the knowledge within their competencies. Among such actors are the Russian Academy of Sciences (RAS), which looks after research projects under execution in SI RAS, the Federal agency of scientific organizations, which determines how evaluate these projects, and many others. To imitate the differences of competencies of those who develop applied social knowledge, GCOO methods urge them to build up the Hierarchy of Contexts or domains (hereafter HC). Such hierarchy within GCOO methods has two features. First, any domain represents a sphere of conceptual actions, socially sanctioned to a certain actor. Second, knowledge, developed in a domain by an actor, may be borrowed by other actors working in other contexts.
In our example, we can imagine the differentiation of the competencies as following. The actors Nancy Smith, John Lincoln and Bill Harrison (hereafter called explicit ones) are employees of SI RAS. Their personal profiles at the level of SI RAS inevitably use notions and formulations put forward by several other institutions (e.g. mentioned above). Let us call these institutions external actors. Both external and explicit actors carry out the conceptualization of the idea of SI RAS. Say, Nancy Smith is characterized by concepts of Social&Demographic Characteristics, Gender, Age, etc. (see Figure 4). These notions were elaborated without any connection to the task of creating a profile of this SI RAS employee, but are to be interpreted through evidence observable at the level of this organization.
To indicate external domains, which predefine many attributes used in profiles, we carry out two operations. First, we introduce corresponding concepts that describe the domains where external actors operate - Federal Agency of Science Organizations, Saint Petersburg, etc. - into the dictionary of the CM. Next, we bind the introduced concepts by creating corresponding branchings, gathered in the thesaurus (see Appendix B).
As a result, GCOO methods algorithms become able to automatically generate the hierarchy of contexts demonstrated in Fig. 8.
. NUMBERS a • COMMON KNOWLEDGE B RUSSIAN FEDERATION B * SAINT-PETERSBURG
• NUMBERS
B . COMMON KNOWLEDGE L-J * RUSSIAN FEDERATION B RUSSIAN ACADEMY OF SCIENCES
• NUMBERS
B • COMMON KNOWLEDGE B * RUSSIAN FEDERATION B 5*? MINISTERY OF EDUCATION AND SCIENCE
• NUMBERS
B . COMMON KNOWLEDGE B W RUSSIAN FEDERATION B » FEDERAL AGENCY OF SCIENCE ORGANIZATIONS B COLLECTIVELY CREATED KNOWLEDGE B 3ft
Figure 8. Hierarchy of competence contexts
On the one hand, this hierarchy represents a differentiation of competencies, socially attributed to the external actors. On the other hand, it represents the relationship of visibility, well known in programming. The structure demonstrated in Fig. 8 shows the visibility of terms with reference to contexts where the terms are defined by corresponding
branchings. The hierarchy of contexts permits us to define contexts for concepts used as attributes in the profiles of explicit actors. For instance, we can define So-cial&Demographic Characteristics in the context Russian Federation, bearing in mind that a socially accepted description of an actor is approved at the level of the state. At the same time, we suggest that ideas of Female and Male are wide spread in communicative practice and used easily by people. Therefore, such ideas do not need any further explanation and we define them in the context Common Knowledge.
Our next suggestion relates concept Official Salary to the competence of regional authorities. Within CM we denote them as Saint Petersburg. We continue this process of conceptual definitions of actor profiles' attributes until all concepts, needed to describe profiles, are present.
Our ontoeditor also allows automatic creation of other geometric views of the generated knowledge, For instance, Fig. 9 reproduces the Fig. 8, but with two distinctions: (1) one concept relates to one node instead of many nodes, and (2) nodes can be connected with many edges instead of only one edge.
Figure 9. Geometric view of hierarchy of competence contexts
Numeric attributes of actors' profiles. Actor profiles can include attributes of different conventional types. For instance, the age assumes measuring in integer or date formats. To obtain such numerical information from the user, the GCOO methods assume that the person will represent the numbers as numerals, making these numerals the names (captions) of the concepts. So the user acquires a possibility to present numeric attributes of his profile. But concepts with numerals as captions are similar to all other concepts gathered in thesaurus. To become operable by GCOO algorithms they must be associated to corresponding contexts. We define all concepts designed to express numeric values in the context expressed by a concept Numbers. This concept is one of designators pre-
sumed to define notions considered generally accepted. Another such context is Common Knowledge, where we gather all non-numeric concepts that do not need further explaining. Figure 8 demonstrates the relationship between Numbers and Common Knowledge.
External and dynamic contexts for describing actors. Let us note, that HC does not contain all contexts used for defining profiles' concepts. The notions Gender, Age and some others exist in contexts that are outside HC. The formulas (2) and (3) show definitions of Gender using contexts Nancy Smith and John Lincoln that are absent in HC (Fig. 8).
(Gender, Nancy Smith)^{(Female ,nothing} (2)
(Gender, John Lincoln)^{(Male, nothing)} (3)
To process such branchings, GCOO methods algorithms apply additional conventions as compared with the above rule f. The conventions extend visibility of the terms used in profiles, as opposed to the rules established only by means of HC. When calculating the children of a certain node of terminological graph, they take as contexts those concepts that were processed and incorporated into the already built part of the graph.
Let us look at how the conventions function during generating terminological graph shown in Fig. 4. For instance, the concept Gender enters in two branches that can be identified by pairs (Nancy Smith, SI RAS) and (John Lincoln, SI RAS). According to (2) and (3), Gender has two possible contexts, so which context should be taken? To pair Gender on each branch with appropriate context (as shown in Fig. 4) the algorithm traces a path that leads from the node under construction with still undefined context (Gender, ?) up to root node (Actual list of employers, SI RAS). Since one of the two existing paths includes the term Nancy Smith and another one contains John Lincoln the algorithm chooses only one context for Gender according to a branch on which the forming node is located.
The conventions enhance the polymorphism of concepts within GCOO methods. The same notion, say Gender, acquires different senses depending on the context where it exists. Fig. 4 demonstrates that Gender with reference to Nancy Smith is Female, and with reference to John Lincoln is Male. The same polymorphic mechanism works in forming all actor profiles: for instance, (Gender, Bill Harrison) accepts attribute Male. In particular, when processed, numeric characteristics of profile also answer to the polymorphic conventions.
4. Automation of describing actors
GCOO methods have various means for simplifying user work needed to create and maintain various structural descriptions of actors. One of such instruments is a tool of assimilation. It provides the use of existing conceptual structures to define with their help new concepts appearing during conceptualization. Let us explain the application of this instrument in the frame of the example shown in Fig. 4. Suppose we made the structural
profile of (Nancy Smith, SI RAS) and believe that Bill Harrison is similar to (Nancy Smith, SI RAS). In this case, we can express this similarity by addressing to special predefined concepts that describe Bill Harrison within a corresponding branching.
The formula (4) shows how the assimilation represented as a branching may look.
(Bill Harrison, SI RAS)^{(Structurally Similar, Nancy Smith); (In Context, SI RAS)} (4)
The formula (4) uses two special concepts Structurally Similar and In Context. The GCOO methods algorithms recognize these concepts and process them in a specific way. Concept Structurally Similar attempts to read and retain the underlying structural definitions of the already existing head pair (Nancy Smith, ... ). In Context works if both the existing pair and the pair under construction share the same context (here..., SI RAS). In particular, it retrieves only the relevant portion of the branchings associated with Nancy Smith, i.e. only those built under context (Nancy Smith, SI RAS). In our case, a description already built up for Nancy Smith will be retained, duplicated and attributed to Bill Harrison.
When generating the terminological graph on the basis of formula (4), the GCOO methods algorithms replaces the special explanations of (4), i. e. those containing reserved concepts Structurally Similar and In Context, with a set of explanations associated in the thesaurus with (Nancy Smith, SI RAS). As a result, the pair (Bill Harrison, SI RAS) gets new explanations borrowed from the definition of (Nancy Smith, SI RAS).
These borrowed explanations are subject to further coordination because of a change of contexts. The initial context for these explanations contained Nancy Smith, and now Bill Harrison has appeared instead. Consequently, polymorphic concepts defined in the both profiles will change their structural views (see Fig. 4). Say, (Gender, Nancy Smith), defined as (Female, Common Knowledge) becomes automatically (Gender, Bill Harrison), defined as (Male, Common Knowledge). The concept Age, which had a value of 40 years in actor Nancy Smith, acquires value of 70 years as it becomes associated with Bill Harrison (to shorten demonstrating materials, we omit concepts that would denote a unit of measurement, for instance, Years).
The reader can follow the presented automatic modifications on his/her own with the help of the graph shown in Fig. 4. Let us note that GCOO methods do not restrict the complexity of conceptual structures employed by the user to describe social actors.
5. Conclusion
Using a practical example, we demonstrated how the GCOO methods allow non-specialists in computer science to create and maintain practical knowledge about social actors. In this respect, the procedures for building and exchanging practical knowledge based on GCOO methods can become an alternative to using Wikipedia as a resource of
accumulating people's experience [18]. What are the possible pros and cons of this replacement is a subject of another study.
In fact, the GCOO methods provide an original graphical way of presenting semantics hidden in the text. Therefore, their usability depends on to which extent the graphical apparatus for redefining the text is clear and intuitively understandable to the natural language speaker. Of course, examples above cannot exhaustively answer this question by themselves. We see the need for further analysis and development of GCOO methods as a tool used by human beings to express text semantics collectively.
Examples given above show that at this phase the user can introduce the numeric characteristics of social objects under conceptualization. Yet they are not available for full-fledged numeric processing. The next step in improving GCOO methods is the development of the instruments for "structural calculations" which will expand the use of OOP capabilities in GCOO methods.
6. Acknowledgements
We thank Yuta Tamberg (Saint-Petersburg State University) for clarifying discussion and editing. Gratitude is also due to the anonymous referees who made valuable comments on the earlier versions of the manuscript.
Referenses
[1] Castells M. (1996) The Rise of the Network Society, The Information Age: Economy, Society and Culture. Vol. I. Cambridge, MA, Oxford, UK, Blackwell.
[2] Doan A., Ramakrishnan R., Halevy A. (2011) Communications of the ACM, April. 54(4):86-96.
[3] WeisfeldM. (2009) The Object-Oriented Thought Process. 3th ed. Addison-Wesley.
[4] Mangabeira W C. (1995) Computer Assistance, Qualitative Analysis and Model Building. In Information Technology for the Social Scientist. Ed. by R. M. Lee. London: UCL Press. P. 129-46
[5] Gibbs G. R., Susanne F., Wilma C M.. (2002) The Use of New Technology in Qualitative Research. Introduction to Iss. 3(2) of FQS. In Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 3(2). http://www.qualitative-research.net/fqs-texte/2-02/2-02hrsg-e.htm
[6] Kanygin G. V., PoltinnikovaM. S. (2016) SPIIRASProceedings, 5(48):107-124. [In Rus]
[7] Poltinnikova M., Kanygin G. (2016). Graphic ontological methods to compile collective knowledge of social processes. In Proc. of the International Conference on Electronic Governance and Open Society: Challenges in Eurasia. ACM. P. 223-228.
[8] Gruber Th. R. (1993) Knowledge Acquisition, 5(2):199-220.
[9] Mylopoulos J. (1992) Conceptual Modeling, Databases, and CASE: An Integrated View of Information Systems Development; chapter Conceptual Modeling and Telos. John Wiley & Sons, Inc. P. 49-68.
[10] Berger P., Lukman T. (1966) Social Construction of Reality. A treatise on the sociology of knowledge. Garden Sity, NY: Anchor Books.
[11] Weber M. Weber einige Kategorien der verstehenden Soziologie. (1913) Logos 4(3):253-94.
[12] Wagner G., Almeida J. P. A., Guizzardi R. S. S. (2015) Applied Ontology 10(3-4):259-271.
[13] Kelle U. (1997) Sociological Research Online. 2(2) http://www.socresonline.org.uk/ 2/2/1.html
[14] Tesch R. (1990) Qualitative Research: Analysis types and software tools. New York: Taylor & Francis Ltd.
[15] Lewins A., Silver C. (2007) Using Qualitative Software: A Step-by-Step Guide. London: Sage publications.
[16] Charmaz K. (2000) Grounded Theory: Objectivist and Constructive Methods. In Handbook of Qualitative Research, Norman K. Denzin & Yvonna S. Lincoln (Eds.), 2nd ed. Thousand Oaks, Ca.: Sage. P. 509-535.
[17] Bringer J. D., Johnston L. H., Brackenridge C. H. (2006) Field Methods, 18:245-266
[18] Wagner C. (2006) Information Resources Management Journal, 19(1):70-83.
Appendix A
Dictionary T={ 1200; 12000; 19000; 32000; 37; 40; 5500; 62500; 70; Actual list of employers; Administrative Status; Age; Bill Harrison; Candidate of Science; Common Knowledge; Doctor of Science; Education; Additional Income; Female; Gender; Higher; Income; John Lincoln; Male; Minor Researcher; Nancy Smith; Numbers; Official Salary; Personal Friend; Principal Researcher; Prosecuter; Russian Federation; Saint Petersburg; Salary; Senior Researcher; SI RAS; Social Status; Social&Demographic Characteristics; Unofficial Income; VIP}
Appendix B
Branchings:
1. (Actual list of employers, SI RAS)^{(Nancy Smith, SI RAS);(John Lincoln, SI RAS);(Bill Harrison, SI RAS)};
2. (Nancy Smith , SI RAS)^-{(Social&Demographic Characteristics, Russian Federation)};
3. (Social&Demographic Characteristics, Russian Federation)^-{(Gender, Nancy Smith);(Age, Nancy Smith); (Social Status, Russian Federation)};
4. (Gender, Nancy Smith)^-{(Female, Common Knowledge)};
5. (Female , Common Knowledge)^—};
6. (Age, Nancy Smith)^{(40, Numbers)};
7. (40, Numbers)^ •••};
8. (Social Status , Russian Federation)^-{(Education, Nancy Smith);(Administrative Status, Nancy Smith); (Income, Russian Federation)};
9. (Social Status , Russian Federation)^-{(Education, Nancy Smith);(Administrative Status, Nancy Smith); (Income, Russian Federation)};
10.(Education, Nancy Smith)^-{(Candidate of Science, Common Knowledge)};
11. (Candidate of Science, Common Knowledge)^—};
12. (Administrative Status, Nancy Smith)^-{(Senior Researcher, Common Knowledge)};
13. (Senior Researcher, Common Knowledge)^—};
14.(Income , Russian Federation)^{(Official Salary, Saint-Petersburg);(Additional Income, Nancy Smith)};
15.(Official Salary, Saint-Petersburg)^-{(Salary, Nancy Smith)};
16.(Salary, Nancy Smith)^{(19000, Numbers)};
17.(19000, Numbers)^ — };
18.(Additional Income, Nancy Smith)^{(1200, Numbers)};
19.(1200, Numbers)^-};
20.(John Lincoln, SI RAS)^{(Gender, John Lincoln); (Age, John Lincoln); (Social Status, Russian Federation); (Social Status, John Lincoln)};
21. (Gender, John Lincoln)^{(Male, Common Knowledge)} ;
22. (Male, Common Knowledge)^ —};
23.(Age, John Lincoln)^{(37, Numbers)};
24.(37, Numbers)^—};
25. (Education, John Lincoln)^{(Higher, Common Knowledge)};
26. (Higher, Common Knowledge)^—};
27. (Administrative Status, John Lincoln)^{(Minor Researcher, Common Knowledge)};
28. (Minor Researcher, Common Knowledge)^ — };
29.(Salary, John Lincoln)^{(12000, Numbers)};
30.(12000, Numbers)^ — };
31. (Unofficial Income, John Lincoln)^{(32000, Numbers)};
32.(32000, Numbers)^ — };
33.(Social Status, John Lincoln)^{(Personal Friend, John Lincoln)};
34.(Personal Friend, John Lincoln)^{(Prosecuter, Saint-Petersburg)};
35.(Prosecuter, Saint-Petersburg)^{(VIP, Common Knowledge)};
36.(VIP, Common Knowledge)^ — };
37.(Bill Harrison , SI RAS)^{(Social&Demographic Characteristics, Russian Federation)};
38.(Gender, Bill Harrison)^{(Male, Common Knowledge)};
39.(Age, Bill Harrison)^{(70, Numbers)};
40.(70, Numbers)^—};
41. (Education, Bill Harrison)^{(Doctor of Science, Common Knowledge)};
42. (Doctor of Science, Common Knowledge)^—};
43. (Administrative Status, Bill Harrison)^{(Principal Researcher, Common Knowledge)};
44.(Principal Researcher, Common Knowledge)^ —};
45. (Salary, Bill Harrison)^{(62500, Numbers)};
46.(62500 // Numbers)^-};
47. (Additional Income, Bill Harrison)^{(5500, Numbers)};
48.(5500, Numbers)^-};
Опыт применения графовых контекстно-ориентированных онтологических методов
Г. В. Каныгин, М. С. Полтинникова
Социологический институт РАН — филиал ФГБУ Науки Федерального Научно-исследовательского Социологического Центра Российской Академии Наук 190005, Санкт-Петербург, ул. 7-я Красноармейская, 25/14 e-mail: g.kanygin@gmail.com, maria.poltinnikova@gmail.com
Аннотация. Авторы предложили графовые контекстно-ориентированные онтологические (ГКОО) методы, которые делают доступными инструменты объектно-ориентированного программирования (ООП) для практического использования неспециалистами в области информатики. В данной статье авторы разъясняют возможности ГКОО методов по масштабированию профилей сотрудников организации. В статье показано, как математическая модель ГКОО методов, реализованная в в виде онторедак-тора, сводится к наглядным пользовательским интерфейсам. Приведена пользовательская методика, которая позволяет неподготовленному человеку описывать социальные процессы и проверять свои естественно-языковые описания с помощью инструментов ООП.
Keywords: ГКОО методы, ветвления, разъяснение, контекст разъяснения, терминологический граф (T-граф), компиляция Т-графа, граф иерархии контекстов.
Литература
[1] Castells M. The Rise of the Network Society, The Information Age: Economy, Society and Culture. Vol. I. — Cambridge, MA; Oxford, UK: Blackwell. 1996.
[2] Doan A., Ramakrishnan R., Halevy A. Crowdsourcing systems on the World-Wide Web // Communications of the ACM. 2011. Vol. 54. No. 4. P. 86-96.
[3] WeisfeldM. The Object-Oriented Thought Process. — 3rd ed. — Addison-Wesley. 2009.
[4] Mangabeira Wilma C. Computer Assistance, Qualitative Analysis and Model Building. In Information Technology for the Social Scientist / Ed. by R. M. Lee. — London: UCL Press. 1995. P. 129-46.
[5] Gibbs, Graham R., Susanne Friese and Wilma C Mangabeira. The Use of New Technology in Qualitative Research. Introduction to Iss. 3(2) of FQS //Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 2002. Vol. 3. No. 2. http://www.qualitative-research.net/fqs-texte/2-02/2-02hrsg-e.htm
[6] Kanygin, G. V., Poltinnikova M. S. Контекстно-ориентированные онтологические методы в социологии // Труды СПИИРАН. 2016. Vol. 5. No. 48. P. 107-124.
[7] Poltinnikova M., Kanygin G. Graphic ontological methods to compile collective knowledge of social processes // Proceedings of the International Conference on Electronic Governance and Open Society: Challenges in Eurasia. — ACM, 2016. P. 223-228.
[8] Gruber Th. R. A Translation Approach to Portable Ontology Specifications // Knowledge Acquisition. 1993. Vol. 5. No. 2. P. 199-220.
[9] Mylopoulos J. Conceptual Modeling and Telos // Chapter in Conceptual Modeling, Databases, and CASE: An Integrated View of Information Systems Development. — Chichester: Wiley John Wiley & Sons, Inc. 1992.
[10] Berger P., Lukman T. Social Construction of Reality. A treatise on the sociology of knowledge. — Garden Sity, NY: Anchor Books, 1966.
[11] WeberM. Weber einige Kategorien der verstehenden Soziologie // Logos. 1913. Vol. 4. No. 3. P. 253294.
[12] Wagner G., Almeida J. P. A., Guizzardi R. S. S. Towards Ontological Foundations for Conceptual Modeling: The Unified Foundational Ontology (UFO) Story // Applied Ontology. 2015. Vol. 10. No. 3-4. P. 259-271.
[13] Kelle U. Theory Building in Qualitative Research and Computer Programs for the Management of Textual Data // Sociological Research Online. 1997. Vol. 2. No. 2. (http://www.socresonline.org.uk/2/2/1.html)
[14] Tesch R. Qualitative Research: Analysis types and software tools. — NY: Taylor & Francis Ltd, 1990.
[15] Lewins A., Silver C. Using Qualitative Software: A Step-by-Step Guide. — London: Sage publications, 2007.
[16] Charmaz K. Grounded Theory: Objectivist and Constructive Methods // Handbook of Qualitative Research / Eds: Norman K. Denzin, Y. S. Lincoln. — 2nd ed. — Thousand Oaks, Ca.: Sage, 2000. P. 509535.
[17] Bringer J. D., Johnston L. H., Brackenridge C. H. Using Computer-Assisted Qualitative Data Analysis Software to Develop a Grounded Theory Project // Field Methods. 2006. Vol. 18. No. 3. P. 245-266.
[18] Wagner C. Breaking the Knowledge Acquisition Bottleneck Through Conversational Knowledge Management // Information Resources Management Journal. 2006. Vol. 19. No. 1. P. 70-83
Авторы:
Геннадий Викторович Каныгин — доктор социологических наук, ведущий научный сотрудник, Социологический институт РАН — филиал ФГБУ Науки Федерального Научно-исследовательского Социологического Центра Российской Академии Наук
Мария Сергеевна Полтинникова — кандидат физико-математических наук, старший научный сотрудник, Социологический институт РАН — филиал ФГБУ Науки Федерального Научно-исследовательского Социологического Центра Российской Академии Наук
