A Tool for Management of Knowledge Dispersed throughout Multiple

References

Carlos S

aenz-Ad

, Francisco J. Garc

ıa-Izquierdo

Angel Luis Rubio

Eduardo S

aenz-de-Cabez

on Irigaray

, Emilio Rodr

ıguez-Priego

and Oscar D

ıaz

Departamento de Matem

aticas y Computaci

on, Universidad de La Rioja, Logro

no, Spain

Departamento de Lenguajes y Sistemas Inform

aticos, Universidad del Pa

ıs Vasco, San Sebasti

an, Spain

Keywords:

Model Analysis, Model Tools, References-enriched Concept Map, Natural Language Processing.

Abstract:

When modeling tasks are performed, it is important that the different modeling team members share a com-

mon vocabulary. This implies not only agreement on the terminology itself but especially on the meaning of

the terms used. To this end it comes in handy to have graphical tools for sharing and analyzing the knowledge

dispersed throughout different sources. In this paper we present RCMTool, a tool for creating References-

enriched Concept Maps (RCM). This technique has been specially designed to facilitate the compact presen-

tation and comparison of different deﬁnitions provided by multiple authors in diverse sources. This paper

presents the main features of RCMTool, based on the development of an RCM metamodel and the inclusion

of a natural language processing engine.

1 INTRODUCTION

A clear delimitation of the particular nounances in the

work context is essential for the correct development

of any modeling task, especially when this task is car-

ried out jointly by teams of medium or large size.

However, the available information can be so broad

that the speciﬁcation of the context often becomes a

complex task in itself. In particular, the simple act

of determining a vocabulary whereby the deﬁnition

of terms is accepted by the whole team is an impor-

tant task. The use of dictionaries, thesauri and ency-

clopaedias (Wikipedia being a case of note) is helpful

but it does not solve the problem. For instance, in pre-

vious experiences related to the development of a tool

for business project management, the authors had to

make a great effort to reach an agreement about what

was the meaning and features of the term “business

process” for the team. The problem is even more pro-

nounced in the ﬁeld of research, in which different pa-

pers may have different and even contradictory deﬁ-

nitions of the same concept. Typically, not all of them

are equally relevant to determine the meaning of the

discussed term. The author prestige or the inﬂuence

of one of the deﬁnitions in the literature makes some

deﬁntions stand out from other, which may make the

nouenances of the meaning of the term are closer to

one or the other.

One way to alleviate this problem is to provide

tools that allow a visual and compact representation of

multiple deﬁnitions. In this paper we present RCM-

Tool, a tool that aims to perform this task either in-

dividually or collaboratively by means of the imple-

mentation of the References-enriched Concept Map

(RCM) technique (Rodriguez-Priego et al., 2013).

In (Rodriguez-Priego et al., 2013) RCM is deﬁned

as a Concept Map that “allows the visualization of a

set of deﬁnitions about a term discussed in the liter-

ature, facilitating the analysis of such deﬁnitions in

relation to the authors who propose them. The appear-

ance of an RCM resembles a Concept Map since it is

basically a diagram showing relationships (links) be-

tween concepts. Nevertheless, RCMs enrich the links

with references to the publications”.

An RCM allows the graphical comparison of dif-

ferent propositions through the Path feature. Fig. 1

depicts a portion of the RCM from Fig. 5 to il-

lustrate its components. In this ﬁgure we can see

four different deﬁnitions of “Language” from differ-

ent sources. For instance, one of this deﬁnitions corre-

sponds to the Path “Language is a system of signs ex-

pressing ideas” (highlighted in the ﬁgure). This Path

is a sequence of Concepts such as “language”(main

concept), “system”, “signs” and “ideas”, graphically

represented with blue boxes. These Concepts are

connected by Linking Phrases -words between blue

Saénz-Adán C., García-Izquierdo F., Rubio Á., Irigaray E., Rodríguez-Priego E. and Díaz O..

A Tool for Management of Knowledge Dispersed throughout Multiple References.

DOI: 10.5220/0005499700790086

In Proceedings of the 10th International Conference on Software Paradigm Trends (ICSOFT-PT-2015), pages 79-86

ISBN: 978-989-758-115-1

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

boxes (“is”, “of” and “expressing”)-. A Path must al-

ways be associated with a corresponding bibliograph-

ical Reference, so that the source of the information

can be traced down (in the example this is the ref-

erence labeled with the number 3). In addition, an

RCM allows to have Concepts to be organized in Lay-

ers (boxes with different shades of blue), so that the

creator of the diagram assigns degrees of relevance to

the Concepts, depending for example on the proxim-

ity of each concept in the map to the main concept

whose meaning he/she is attempting to clarify. All of

these features ensure that an RCM is a type of model

that can represent knowledge gathered from deﬁni-

tions that are dispersed throughout multiple sources.

The RCMTool pursues three objectives. First, it

allows the management of bibliographical references.

These sources can be inserted manually, also through

ﬁles (in BIBT

X format) or through integration with

reference management tools like Mendeley. Sec-

ond, RCMTool facilitates the graphical representation

of deﬁnitions through paths. A Path can be manu-

ally constructed (graphically with building blocks of

RCM) or created in a semi-automated way by a re-

stricted natural language processing assistant. Finally,

the tool should allow collaborative use, enabling sev-

eral creators of an RCM to work on the same diagram

providing new deﬁnitions or references or improving

those already existing.

RCMTool is a web tool developed using consol-

idated technologies such as HTML5/CSS3, NodeJS,

JS, oAuth, and REST APIs. It uses Rappid (Client

IO, 2014), a toolkit for building interactive diagram-

ming applications, under an Academic User License.

The paper is structured as follows. Section 2

shows the related work. Sections 3 and 4 explain the

fundamentals and features of the tool. RCMTool at

work is shown in Section 5. The last section sets out

our conclusions and further work.

2 RELATED WORK

Various techniques for knowledge representation ex-

ist in the literature. Many of these techniques are

based on the graphical representation of ideas, con-

cepts, topics or terms (in many cases using graphs).

Topic Maps (Park and Hunting, 2003) and Mind

Maps (Buzan, 2006) are two cases in point. Mind

Maps has tools, such as FreeMind (FreeMind Dev.

Team, 2014) or XMind (XMind Ltd., 2014) that al-

low the computer-based creation of that kind of maps.

On the other hand, Concept Maps (Novak and Ca

nas,

2006) stands out for its extensive use. Concept Maps

were created by Novak, based on Ausubel’s theory of

meaningful learning (Ausubel, 1963; Ausubel et al.,

1968).

When developing modeling tasks, however, often

what is important is not only the representation of

knowledge but a strong consensus on the concepts and

terms used. This consensus facilitates the modeling

task since ambiguities and misunderstandings among

members of the development team are avoided. We

found that although it might be feasible to use Con-

cept Maps (or other techniques) in order to compare

and to present deﬁnitions in a compact way, none

of the existing techniques are perfectly suited to this

task. On the one hand, the goal of a Concept Map is

not to present deﬁnitions of concepts, but to represent

knowledge about concepts. On the other hand, a Con-

cept Map does not facilitate comparison of different

deﬁnitions available from various sources, nor does it

have a precise system for managing these sources.

Inspired by Concept Maps, the References-

enriched Concept Maps (RCM) tech-

nique (Rodriguez-Priego et al., 2013) has been

developed to include several features, mainly path

labeling and concept layering, which meet the needs

explained above. The RCM technique also includes

additional features to measure the complexity of

the included deﬁnitions and their relationship with

their corresponding references. These features allow

authorship attribution to be mantained. This infor-

mation, in turn, allows the RCM creator to consider

the relative weight of different deﬁnitions (e.g. a

deﬁnition included in an online encyclopaedia could

be considered less relevant than another included

in an article in a prestigious scientiﬁc journal),

representing a certain kind of knowledge modeling.

CMapTool is also of interest (Ca

nas et al., 2005).

It focuses on Concept Map, emphasizing its informa-

tion visualization features. This tool is not tailored

to the speciﬁc features of RCM. It allows the associ-

ation of information resources to concepts or linking

words on the map, thanks to which a Concept Map

can be used as a repository of information about a

certain topic. However, it does not allow the associa-

tion of resources to sequences of concepts and linking

phrases (paths). Since the above feature is not avail-

able in any system, yet of paramount importance, so

it has been necessary to develop a tool that realizes it.

3 TOOL FUNDAMENTALS

The tool has been designed keeping in mind two

fundamental principles. Firstly it clearly distin-

guishes the graphical representation from the con-

ceptual model behind an RCM. In a similar way

ICSOFT-PT2015-10thInternationalConferenceonSoftwareParadigmTrends

Figure 1: Portion of RCM shown in Figure 5.

to (Kowata et al., 2010), who propose a metamodel

for Concept Maps, we have developed an RCM meta-

model. Using this approach, each particular RCM is

a model, since it is an instance of a metamodel. This

instance is linked internally with the graphical repre-

sentation drawn by the tool. Secondly, the tool has

a natural language processing engine. This engine

can largely automate the construction of RCMs, since

when a restricted natural language text deﬁnition is

typed, the tool is able to draw the path corresponding

to that deﬁnition.

3.1 Metamodel

Fig. 2 shows the core of the RCM metamodel. This

metamodel is completed with a series of Object Con-

straint Language (OCL).

The two basic building blocks (RCMNodes) of any

RCM are Concept and LinkingPhrase. With these ba-

sic elements, Paths are constructed as an ordered list

of Concepts connected through LinkingPhrases. In

order to precisely deﬁne the notion of Path required

by RCM it is necessary to include the technical notion

of Syntagm. Thus, a Path is actually a list of Syntagms

(which in turn are formed joining a Concept and a

LinkingPhrase). The semantic construction of Paths

is completed with the following OCL constraints:

context Path

inv: syntagmList->

excluding(syntagmList->first())

->forAll(elements.size()=2)

inv: (syntagmList->first()).elements->

size()=1

inv: (syntagmList->first()).elements.

oclIsKindOf(Concept)

These rules set up the facts that “Every Syntagm

in syntagmList from a Path has two elements except

the ﬁrst” and that “The ﬁrst Syntagm in syntagmList

has only one element, and this element is a Concept”.

Other relevant properties are represented in the

RCM metamodel. Both Concepts and LinkingPhrases

may be included in several different Paths. This prop-

erty enables the comparison of deﬁnitions collected

from different sources, by calculating metrics of sim-

ilarity between groups of deﬁnitions. Furthermore,

Concepts are assigned to Layers, following the crite-

ria chosen by the RCM creator. Finally, every Path

has a Reference associated with it, and one Reference

may include several deﬁnitions (and therefore have

several associated Paths).

3.2 Natural Language Processing

According to (Novak, 1984) it is essential to iso-

late concepts from linking words (LinkingPhrases in

RCM). Both are important as language units, but they

play different roles in the transmission of meaning.

While concepts are labeled by words that represent

things and events, relationships are a way of linking

two concepts in a propositional form.

With the aim of processing the mark up of a

word in a text as corresponding to a particular part

of speech, based on both its lexicon, as well as its

context, we use part-of-speech tagging (POS tagging

or POST). This technique (also called grammatical

tagging or word-category disambiguation) has been

used for tagging words in accordance with the part

of the speech tag-set used by the Penn Treebank

Project (Santorini, 1990) (see Table 1).

We could consider Noun Phrase (NP), Verb Phrase

(VP) and Prepositional Phrase (PP) as primary can-

didates to be mapped as RCMNodes. Noun Phrase

(NP) is the most important candidate to be a concept;

Verb Phrase (VP) is aspirant to be a LinkingPhrase,

and Prepositional Phrase (PP) could be both a con-

cept, taking the essence of the Noun Phrase, and a

relationship, if we take into consideration the prepo-

sition that appears before the Noun Phrase.

The implemented algorithm is based on (Brill,

1995), and it is used to apply transformation rules

(some examples of these rules are shown in Table 2).

This algorithm tags words using the rules and follow-

ing the schema in Fig. 3. The original, unannotated

text moves to a second state called “initial state”. In

this second state words are tagged in four steps which

are independent of the context: ﬁrst, tag words from

AToolforManagementofKnowledgeDispersedthroughoutMultipleReferences

Figure 2: RCM Metamodel.

Table 1: List of tags of corresponding part of speech used by Penn Treebank project.

Number tag Description Number tag Description

1 CC Coordinating conjunction 2 CD Cardinal number

3 DT Determiner 4 EX Existencial there

5 FW Foreign word 6 IN Preposition or subordinat-

ing conjunction

7 JJ Adjective 8 JJR Adjective, comparative

9 JJS Adjective, superlative 10 LS List item marker

11 MD Modal 12 NN Noun, singular or mass

13 NNS Noun, plural 14 NNP Proper noun, singular

15 NNP Proper noun, plural 16 PDT Predeterminer

17 POS Possesive ending 18 PRP Personal pronoun

19 PRP$ Possesive pronoun 20 RB Adverb

21 RBR Adverb, comparative 22 RBS Adverb, superlative

23 RP Particle 24 SYM Symbol

25 TO to 26 UH Interjection

27 VB Verb, base form 28 VBD Verb, past tense

29 VBG Verb, gerund or present

participle

30 VBN Verb, past participle

31 VBP Verb, non-3rd person sin-

gular present

32 VBZ Verb, 3rd person singular

present

33 WDT Wh-determiner 34 WP Wh-pronoun

35 WP$ Possessive Wh-pronoun 36 WRB Wh-adverb

a dictionary; second, tag words from a sufﬁx; third,

tag words as NN if they are not the ﬁrst word and

they are capitalised; and ﬁnally transform numbers

and tag as CD. After this step the text is annotated,

although some errors can occur. These errors are cor-

rected with transformation rules (Table 2) which are

context-dependent.

Example. Consider the statement “Language is

a set of sentences”. This proposition has been frag-

mented and tagged in the initial state as follows :

• Language/ NN

• is /VB

• a /DT

• set /VB

• of /IN

• sentences /NNS

ICSOFT-PT2015-10thInternationalConferenceonSoftwareParadigmTrends

Figure 3: Flow for tagging based on (Brill, 1995).

Table 2: Some transformation rules.

From To Condition

- NN Before a MD

NN VB After a RB

JJ RB Before a JJ

VB NN After a DT

VB JJ After CP and RB

VB JJ After DT before NN

The tagging task is performed without context and

has errors, since the word “set” has been tagged as

VB despite the fact that in this example “set” means

a number or combination of things of similar nature

or function, and is therefore a noun. This error is cor-

rected by means of the application of transformation

rules in the last step. There is a rule that states that if

a word is a VB and the preceding word is a DT, the

ﬁrst word will be tagged as NN. Hence, as “set” is

preceded by a DT it has been tagged to NN.

4 TOOL DESIGN

The design of the tool incorporates aspects related

both to functionality and the user interface.

4.1 Features

RCMTool was developed with several key features

(Fig. 4) in mind. Naturally, and since it is a di-

agramming tool, addition and deletion of the build-

ing blocks of RCM (namely Concepts and Linking

phrases) and editing of their properties are allowed.

Linking of elements is no longer immaterial, since

only concepts with linking phrases (in either direc-

tion) must be connectable. It makes no sense that two

concepts (or linking phrases) are connected to each

other, since by deﬁnition of RCM must be possible

traverse a path in a natural way and read a meaning-

Figure 4: Functionality of the tool.

ful statement. Furthermore, the user must be able to

edit link structure’s and labels, which are lists of ref-

erence identiﬁers. One of the key features of RCM

is the creation of paths. Several elements (both con-

cepts and linking phrases) can be selected simultane-

ously and can be then marked as a path. This path

corresponds to a statement that appears in a biblio-

graphical reference, so that both path and reference

are related. The computation of the metrics for de-

termining the complexity of an RCM and the collab-

orative use of the tool are other features that should

be included. Finally, it is important that the user can

export RCMs to other formats, both as graphics (such

as JPEG) and other formats based on data processing

(such as JSON). This export feature facilitates con-

nectivity and interoperability with other tools. Thus

RCM can be incorporated into other documents or an-

alyzed by speciﬁc tools.

4.2 User interface

The interface is split into areas in line with the func-

tionality of the tool (see Fig. 5).

1. Toolbar. Located at the top of the interface. It al-

lows users to add anonymous propositions, add a

list of propositions with their corresponding ref-

erences, add new references to the references list,

highlight a path, import/export the RCM, clear the

canvas and create an image of the RCM.

2. Stencil. Located on the top left of the interface

and groups the main building blocks of an RCM

(Concepts and LinkingPhrases). The user can add

them by dragging them onto the workspace.

AToolforManagementofKnowledgeDispersedthroughoutMultipleReferences

3. References list. Located at the left of the interface

and shows the available references. Each refer-

ence has an identiﬁer, a list of authors and the title

of the article or the URL.

4. Paths list. Located on the bottom left of the inter-

face. It shows the existing paths, each path has an

identiﬁer linked with the corresponding reference.

5. Workspace. it is the main editing area, located in

the centre of the interface.

6. RCMs list. Located at the top right of the interface

and shows all the RCMs shared by the user.

7. Inspector. Located on the right of the interface. It

is used to display and edit properties of elements.

8. Status bar. Located at the bottom of the interface.

It displays information about the application state.

5 TOOL USAGE

The tool has been developed as a web application,

based on the metamodel explained in Section 3.1,

implementing the features listed in Section 4.1 and

following the user interface design described in Sec-

tion 4.2. The tool also includes a natural language

processing engine that uses the algorithm described

in Section 3.2.

5.1 Drawing Interaction

There are three different ways to add elements in or-

der to build an RCM:

• The user can create a branch of the graph through

an anonymous (i.e. without an associated refer-

ence) textual statement.

• Branches of the graph can be created through a

text containing statements associated with their

references.

• The user can add new elements by manually drag-

ging them from the stencil to the workspace and

then editing the text property.

In the two ﬁrst cases, the tool uses the natural lan-

guage processing engine (described in Section 3.2).

Although this process splits sentences into concepts

and linking phrases, it does not generate a ﬁnal re-

sult in all cases, but provides powerful support for

the construction of the graph, which can be manually

edited by the RCM creator. It is important to note that,

since each concept is unique and all the elements in

the graph must be associated with at least one refer-

ence (see Fig. 2), the elements that temporarily have

no associated reference are depicted in red to serve as

an aid during construction.

The tool generates the graph following RCM con-

struction principles, and therefore it will be a directed

graph with only one main concept on the left side of

the graph. The rest of the concepts are automatically

located on the right of the main concept following the

statement structure. When, in the opinion of the cre-

ator, the RCM is structurally complete, the user can

assign layers to the included concepts.

5.2 References and Paths Interaction

One of the main features in RCMs is references man-

agement. Every deﬁnition and every path must have

an associated reference. The graph reﬂects the fact

that different paths can share concepts or linking

phrases, so directed links between elements are la-

beled with the applicable references (see example in

Fig. 5).

The tool manages a repository of references that

the user can add in three different ways. Firstly, the

user can add a reference manually in BIBT

X format.

Secondly, the references may come from a BIBT

ﬁle. Lastly, references can be imported from exter-

nal repositories such as Mendeley (which requires a

user account). OAuth 2.0 authentication is used in or-

der that the user can give secure access to their saved

references in Mendeley. This is an open authorisa-

tion standard, which provides client applications like

RCMTool secure delegated access to server resources

on behalf of a resource owner.

Each of the references in the RCMTool internal

repository is identiﬁed by a number, and includes ba-

sic information about the source: title, authors, date,

URL, etc. Whenever a reference is assigned to a set

of correlative elements, one path is created. This is

the case when branches of the graph are created from

referenced statements by using the engine (see previ-

ous subsection). Paths must have a unique identiﬁer

that links them to the reference (reference identiﬁer).

If two paths are connected to the same reference (e.g.,

when the same source includes two different deﬁni-

tions of the same concept), the tool disambiguates

them with a different label, such as the reference iden-

tiﬁer and a letter of the alphabet ( i.e. 1, 1b, 1c, etc.).

5.3 Example

Fig. 5 shows the tool in use with an RCM collect-

ing deﬁnitions of the ‘Language’ concept. Since this

is a highly complex concept, for the sake of brevity

we present an RCM that includes only ten deﬁnitions

from nine different references.

ICSOFT-PT2015-10thInternationalConferenceonSoftwareParadigmTrends

Figure 5: Tool capture with the RCM of ‘Language’. The areas marked with large red numbers correspond to the list in

Section 4.2.

Figure 6: References list (zoomed).

Following the procedures described in the previ-

ous sections, we have included the set of references,

in this case by connecting the tool to Mendeley. The

references list appears in area 3 of the interface, en-

larged in Fig. 6 for better readability. The deﬁnitions

that we want to represent are extracted from these ref-

erences (each number in the following numbered list

corresponds to the reference identiﬁer in Fig. 6).

1. Language is a system of symbols based on con-

vention.

2. Language is a system of codes.

3. Language is a system of signs expressing ideas.

4. Language is a set of expressions used to commu-

nicate.

5. Language is a set of strings from alphabet.

6. Language is a set of system.

7. Language is a set of sentences constructed of ﬁ-

nite set of elements.

8. (a)Language is a set of sentences. (b)Language

has speciﬁc rules.

9. Language is a concept.

These deﬁnitions are included in a list of state-

ments, including a reference identiﬁer for each state-

ment. The tool processes the list, and generates a

graph that can be revised for corrections. In the last

step, the user must classify concepts in the four avail-

able layers (depicted with a color gradation in the

graph).

6 CONCLUSIONS AND FURTHER

WORK

We have presented RCMTool, a web application ded-

icated to the creation of References-enriched Concept

AToolforManagementofKnowledgeDispersedthroughoutMultipleReferences

Maps. This technique allows to the compact repre-

sentation of multiple deﬁnitions of the same concept,

and the comparison of these deﬁnitions by using dif-

ferent criteria and metrics. These features are espe-

cially useful when modeling tasks are performed, in

which there are often ambiguous concepts whose def-

inition must be agreed by all members of the develop-

ment team. Additionally, RCMs themselves are mod-

els, since they are created as instances of the meta-

model that presented in Section 3.1. In this sense

the RCM technique acts as a modeling language that

allows RCM modeling, and therefore RCMTool is a

tool that enables the creation of such models.

RCMTool is a web application ready for further

expansion. Among future improvements we will con-

sider, for example, the automated computation of

metrics (Rodriguez-Priego et al., 2013), and the inclu-

sion of an advanced authoring system that facilitates

the collaborative use of the tool.

ACKNOWLEDGEMENTS

This work was partially funded by Project

MTM2014-54151-P.

REFERENCES

Ausubel, D. P. (1963). The psychology of meaningful verbal

learning. Grune & Stratton.

Ausubel, D. P., Novak, J. D., Hanesian, H., et al. (1968). Ed-

ucational psychology: A cognitive view. Holt, Rine-

hart and Winston New York.

Brill, E. (1995). Transformation-based error-driven learning

and natural language processing: A case study in part-

of-speech tagging. Comput. Linguist., 21(4):543–565.

Buzan, T. (2006). Use Your Head. Mind Set Series. BBC

Active.

nas, A. J., Carff, R., Hill, G., Carvalho, M., Arguedas,

M., Eskridge, T. C., Lott, J., and Carvajal, R. (2005).

Concept maps: Integrating knowledge and informa-

tion visualization. In Knowledge and information vi-

sualization, pages 205–219. Springer.

Client IO (2014). Rappid toolkit. http://jointjs.com/about-

rappid. [Stable release: Rappid 1.4 / September 16,

2014].

FreeMind Dev. Team (2014). FreeMind tool.

http://freemind.sourceforge.net/wiki/index.php/Main

Page/. [Stable release: FreeMind 1.0.1 / April 12,

2014].

Kowata, J. H., Cury, D., and Boeres, M. C. S. (2010). Con-

cept maps core elements candidates recognition from

text. In Proceedings of Fourth International Confer-

ence on Concept Mapping, pages 120–127, Vi

na del

Mar, Chile.

Novak, J. D. (1984). Learning how to learn. Cambridge

University Press.

Novak, J. D. and Ca

nas, A. J. (2006). The theory underlying

concept maps and how to construct them. Technical

report, Institute for Human and Machine Cognition,

Pensacola Fl.

Park, J. and Hunting, S. (2003). XML Topic Maps: Creating

and Using Topic Maps for the Web. Addison-Wesley.

Rodriguez-Priego, E., Garcia-Izquierdo, F. J., and Rubio,

A. L. (2013). References-enriched Concept Map: a

tool for collecting and comparing disparate deﬁnitions

appearing in multiple references. Journal of Informa-

tion Science, 39(6):789–804.

Santorini, B. (1990). Part-of-speech tagging guidelines for

the Penn Treebank Project (3rd revision). Technical

report, University of Pennsylvania.

XMind Ltd. (2014). XMind tool. http://www.xmind.net/.

[Stable release: XMind 6 (v3.5.1) / 20th, Nov 2014].

ICSOFT-PT2015-10thInternationalConferenceonSoftwareParadigmTrends