AUTOMATIC ANALYSIS OF ASYNCHRONOUS DISCUSSIONS

Breno Fabrício Terra Azevedo

, Patricia Alejandra Behar

and Eliseo Berni Reategui

Department of Information Technology, IFF, 273, Dr. Siqueira, Zip Code 28030-130, Campos dos Goytacazes-RJ, Brazil

Graduate Program in Information Technology in Education, UFRGS

110, Paulo Gama, building 12105, 3

floor, room 332, Zip Code 90040-060, Porto Alegre-RS, Brazil

Keywords: Qualitative Analysis, Asynchronous Discussions.

Abstract: This paper presents results of an automatic analysis of text contributions made by students in asynchronous

discussions. The study was carried with the MineraFórum software. Data collected with the program were

compared to appraisals made by teachers. Results show that the average of the analyses of posts made

MineraFórum is similar to the average obtained in the analyses made by teachers.

1 INTRODUCTION

According to Gilbert and Dabbagh (2005), an

important pedagogical benefit of asynchronous

communication is its potential to support the co-

construction of knowledge through discourse.

The study presented by Garrison, Anderson and

Archer (2000) suggests that text-based

communication offers time for reflection. The

authors’ review of the literature indicates that

written communication is closely related to careful

and critical thinking. Writing can be crucial when

the objective is to facilitate thinking over complex

issues, and meaningful and deep learning.

Palloff and Pratt (2004) say asynchronous

discussions must be stimulated by teachers, as they

are the best way to establish interactions among

students. According to the authors, student

interactions provide time for reflecting over studied

educational contents. The ability to reflect is crucial

to virtual students, and should be stimulated.

Discussion forums are a suitable space to offer this

type of action. By participating in the discussion or

simply replying to messages, students indicate that

they are actually reflecting. The authors also

emphasize the importance of the teacher’s role in

discussion forums. Besides writing messages of

support and motivation to students, and answering

their questions, teachers should observe the level of

participation of each learner. In case the teacher

identifies that a student is not participating properly

or digressing from the topic of discussion, he/she

should try to help learners to overcome their

difficulties, and solve problems.

Getting involved in asynchronous discussions,

such as in forums, is an important activity for

students. By analyzing student interactions in

forums, the teacher can diagnose information on

learners. However, if the teacher has a significant

number of students, he/she will need a great amount

of time to do text analysis. A resource that allows

the automatic analysis of posts in discussion forums

can be of great help to teachers. This resource may

allow teachers to identify students who are debating

over the topic of discussion, as well as those who are

not. By doing so, teachers can have extra time to

find out the reasons why some of the learners did not

discuss concepts related to the topic. In case the

teacher identifies students with learning difficulties,

help can be offered.

To perform automatic analysis of texts produced

by students in asynchronous discussion, this paper

presents a study carried with the software

MineraFórum

MineraFórum (Azevedo et al., 2011a; 2011b)

uses text mining techniques to analyze posts in

threaded discussion. By doing this analysis, it is

possible to identify if text contributions produced by

learners are relevant or irrelevant in the debate.

Next section presents a brief introduction to text

mining. Section 3 informs on some works that use

this technique in the analysis of discussion forums.

Section 4 explains the software MineraFórum.

Section 5 describes the experiments, and section 6

presents the concluding remarks.

Terra Azevedo B., Behar P. and Berni Reategui E..

AUTOMATIC ANALYSIS OF ASYNCHRONOUS DISCUSSIONS.

DOI: 10.5220/0003896900050012

In Proceedings of the 4th International Conference on Computer Supported Education (CSEDU-2012), pages 5-12

ISBN: 978-989-8565-06-8

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

2 TEXT MINING

According to Feldman and Sanger (2007), text

mining can be defined as an intensive process of

knowledge in which the user interacts with a great

number of documents by using tools to perform

analysis. The objective is to extract useful

information from a collection of documents. This

information is identified in interesting patterns found

in non-structured text data.

Text mining systems are based on pre-processing

routines, algorithms for discovering patterns, and

elements for presenting results. The system user

interacts with the pre-processing stage, with the

mining nucleus, and with result output.

Pre-processing operations are based on the

identification and extraction of representative

features of documents in natural language. These

operations are responsible for changing non-

structured data, stored in collections of documents,

into a structure expressed in an intermediary model

(Feldman and Sanger, 2007; Tan, 1999). The

intermediary models are based on choice of the

minimum text unit: word, concept, sentence,

paragraph, or document (Torre et al., 2005).

Operations in the mining nucleus, also called

knowledge distillation processes, represent the core

of a text mining system, and involve: pattern

discovery, trend analysis, and incremental

algorithms for knowledge discovery. The most used

mechanisms are distributions and proportions, sets

of frequent concepts, and associations. Activities can

also be related to comparisons, and to the

identification of levels of interest with some patterns

(Feldman and Sanger, 2007).

Elements involved in the presentation of results

represent the system interface, with navigation

function, and access to the language used in the

search (Feldman and Sanger, 2007; Puretskiy et al.,

2010; Tan, 1999).

Text mining explores techniques and

methodologies from areas such as information

retrieval, information extraction, and corpus

linguistics. To extract useful information, one must

discover relevant characteristics in the documents,

the most usual being: characters, words, terms, and

concepts. Characters are individual letters, numbers,

special characters, and spaces. Words are

represented by clusters of characters. Terms are

unique words or sets of words selected straight from

the text. Concepts are features generated for a

document by using different methodologies. Hybrid

approaches can be used to generate document

representation based on features. For example, one

can first extract terms from a text, and then adapt

them by comparing them to a list of relevant topics

(concepts) obtained by categorization (Feldman and

Sanger, 2007).

Considering the four features described above

(characters, words, terms, and concepts), terms and

concepts are those that possess the highest semantic

level. There are many advantages in using those

features to represent documents in text mining.

Representations using terms can be more easily

generated from the original text if compared to

concepts. However, representations with concepts

are better than any other. They can also be processed

in order to support very sophisticated hierarchies by

using knowledge of the domain given by ontologies

and knowledge bases (Feldman and Sanger, 2007).

Text mining using graph technique discovers

words with greater occurrence in texts, and identifies

if they are near one another. The graph obtained in

the mining process presents the most frequent words

in its nodes. Associations between nodes indicate the

proximity between words. Figure 1 represents the

graph generated from the text “There are several

techniques in text mining. Some techniques used in

text mining include: information extraction, topic

tracking, summary production, text categorization,

text clustering, conceptual links, information

visualization, analysis of questions and answers”.

Figure 1: Graph generated from a text.

3 ANALYSIS OF DISCUSSION

FORUMS WITH TEXT MINING

Rebedea et al. (2008) present an analysis of chats

that may be used in threaded discussion. Their study

proposes extraction of socio-semantic data from

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

conversations produced by participants using text

mining techniques based on ontologies. This method

uses a combination of a social-cultural and dialogic

perspective with text processing techniques. The

study also presents the software developed to

discover the most relevant topics in a debate, the

contribution of each participant in the conversation,

and how it can offer a representation of multiple

voices in the conversation. Text mining was used to:

analyze if the content of chat messages is related to

the discussion theme, and determine the moment a

new topic is introduced in the discussion. WordNet

was used to identify synonyms in the texts selected

for the study.

Ravi and Kim’s work (2007) presents an

approach to make automatic identification of

features in students’ posts in threaded discussions.

The authors used word sequence resources and SVM

algorithms (Support Vector Machine) to develop

“speech acts” classifiers to identify purposes in

individual messages, such as: questions, answers,

formulations, corrections. Classifiers were used to

find messages containing questions or answers.

Authors used a set of rules for topic analysis in order

to find out those that could contain unanswered

questions and need the teacher’s attention.

A discussion forum with advanced technological

features is presented by Li et al. (2008). This project

uses domain ontology and text mining techniques. In

this study, transcriptions of discussion forums are

automatically changed into a structural modelling in

three stages: topic acknowledgment, identification of

the type of transcription, and the semantic

association among them. The first step clusters

messages from a set of discussion into a document.

Each document is represented by a vector of

weighted terms. Cosine method is used to calculate

similarity between the vector in the document and

the vector of concepts in the domain ontology. The

second step identifies six types of messages:

question, opinion, suggestion, recommendation,

request, and reference. The third step uses the SLN

model (Semantic Link Network) to organize texts

with semantic association. The forum used in the

study offers three functions to teachers: search of

information considered useful to their needs,

thematic navigation through messages, and

recommendation to students that might be interested

in communicating and collaborating. An experiment

was carried to demonstrate the effectiveness of the

approach to find learning peers with the same

interests, and message search with thematic

navigation.

4 MINERAFÓRUM

MineraFórum is a program that makes qualitative

analysis of posts in discussion forums. It was

developed at NUTED

iii

/PGIE/UFRGS. At present,

version 3.0 of the software is being used. This

program is capable of calculating the relevance of

each post within a particular discussion. To analyze

the content of text contributions, the program uses

the text mining using graph technique. Figure 2

presents the main interface of MineraFórum.

Figure 2: Main Interface of MineraFórum showing

selection of the “File” menu.

Some resources offered by version 3.0 of the

MineraFórum are listed below:

• It allows the user to load or type reference text

on the discussion topic.

• If the user wishes so, instead of informing a text

of reference, it is possible to type concepts

considered to be relevant in the discussion, and

make associations among them.

• It uses a thesaurus in the mining process. This

type of dictionary was previously defined in the

software. Nevertheless, if the user finds it necessary,

another synonyms dictionary can be informed.

Synonyms are important when MineraFórum

compares words typed in the posts with the concepts

considered as relevant in the reference text.

• In addition to the thesaurus, the user can inform

words that are semantically equivalent.

• It calculates the relevance of each post.

• It shows a graphic with the mean relevance level

of messages posted by each author.

• It identifies similar messages written in the

discussion forum.

• It allows results of the mining process to be

stored in html files.

AUTOMATICANALYSISOFASYNCHRONOUSDISCUSSIONS

• It shows a report with information on the

analysis of posts: total number of messages written

by each student, the amount of relevant contribution

made by individual learners, concepts used in the

relevant posts, relevance of each message,

information if the message is similar (or not) in the

forum, the relevance average in posts written by

each student, the number of times each message was

cited in the debate.

Figure 3 shows the MineraFórum interface after the

user selected the button “Mining Forum”.

Informations about the mining are presented: name

of the forum, data of mining, and the total number of

messages posted by all the students. For each

learner, one can see: full name, total of messages,

relevance average of posts. For each message, the

software shows the relevance value and four links

([Info] [Message] [Text concepts] [Forum

concepts]).

• Info: provides information on the number of

times the message was cited in the forum, and if it is

similar to any other. Figure 4 shows informations

about the first post of student 3.

• Message: shows the first characters of the

message. Figure 5 shows the second message of

student 3.

• Text concepts: presents the concepts indicated by

the reference text found in the post.

• Forum concepts: shows the main concepts used

in the forum and found in the posts. Figure 6 shows

the relevant concepts cited in the forum that were

found in third post of student 3.

Figure 3: Interface of the “Mining Forum” window.

During the process of analyzing posts,

MineraFórum organizes and clusters the student’s

messages. The software calculates the relevance

value of each message. To do this procedure, three

criteria are considered: the thematic relevance of the

message (TR), relevance of message reference

(MR), and similarity of the message (MS).

Figure 4: Informations about the first post of student 3.

Figure 5: Second message of student 3.

Figure 6: Relevant concepts cited in the forum that were

found in third post of student 3.

To calculate the thematic relevance of messages

(TR), MineraFórum performs the following actions:

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

a) From the reference text, it builds a graph of the

discussion topic indicated by the user. In this

process, stopwords (words that can be removed in

the mining stage, such as adverbs, articles, and

prepositions) can be deleted, and the most recurring

words in the text are identified. The most frequent

words represent the most relevant concepts in the

mined text, and correspond to the vertices in the

graph. The edges between the vertices are created

according to the proximity between words. If the

user decides to insert important concepts related to

the discussion topic, instead of indicating the

reference text, MineraFórum builds the graph from

those concepts.

b) Automatic loading of all posts. The software

interacts with the Virtual Learning Environment

where the forum took place and get all the messages.

c) Generation of a graph by mining each post.

d) To calculate thematic relevance of a post in

relation to the reference text, MineraFórum analyzes

the correspondence between the generated graph

from the text and graph built from the message. In

this case, one can identify which vertex in the first

graph are equivalent to those in the second graph.

MineraFórum considers that two vertices are

equivalent if they present similar content, that is, (i)

if they have the same words, (ii) if the words can be

reduced to the same stem, (iii) if they have

synonyms, and (iv) if they are semantically

equivalent. In the second stage of analysis, the

program uses a formula that takes into consideration

three aspects of the equivalent vertices: the amount

of such vertices in the two graphs, the distance

between them in the respective graph, and their

weight in their own graphs. The resulting value

corresponds to the thematic relevance of the post in

relation to the reference text (TR

e) Generation of a text graph built from the whole

set of posts, named “forum graph”. The procedure

used to find TR

is used to calculate the thematic

relevance of the post in relation to the forum graph.

The resulting value is named TR

f) Thematic relevance (TR) of a post is calculated

from the average between TR

and TR

To calculate the MR value of a post, the software

divides the number of times the message has been

cited by the total number of posts in a forum. The

computation of the similarity of a message (MS)

with other was held with text mining using graphs.

The graphs of the messages that have similar values

of TR are compared to verify if the posts are similar.

If the post is similar to another in the forum, the MS

value will be equal to TR, with a negative sign.

The relevance value of a post (PR) is obtained

from the weighted mean between TR and MR. If the

message is similar to another, the MS value will be

subtracted from PR. If the text contribution is not

similar to another in the forum, then the calculated

value for PR will be maintained.

The final value for PR is converted into a whole

value, in a scale from 0 (zero) to 5 (five). Zero value

means the message is not relevant in the debate.

Value five indicates that the post has maximum

relevance.

Table 1 presents a comparison between

MineraFórum and the correlational studies presented

in section 3. The comparison shows an analysis of

the similarities and differences between

MineraFórum and other works.

Table 1: Comparison between MineraFórum and

correlational studies.

Authors

Similarities between

MineraFórum and

correlational work

Differences between

MineraFórum and

correlational work

Rebedea et al.

(2008)

Tool that uses text

mining techniques to

analyze relevance of

messages in relation to

the discussion topic.

Analyzes chats

messages to provide

indicators related to

Bakhtin’s theory of

poliphony.

The tool is not

integrated into a VLE.

Ravi and Kim

(2007)

Classifiers to analyze

content of forum

messages as an aid for

teachers.

Classifiers analyze texts

by identifying “speech

acts”.

Indicators given by

classifiers are different

from those presented by

MineraFórum.

Li et al.(2008)

Tool that uses text

mining techniques to

analyze relevance of

posts in relation to the

discussion topic.

The tool represents texts

with vectorial space

model, and identifies if

they are similar by using

the cosine similarity

measure.

The system developed

by the authors offers

help to students, while

MineraFórum presents

information to help

teachers.

MineraFórum is a resource that can be used by

teachers in a Virtual Learning Environment (VLE).

It is integrated into ROODA

(Behar, 2007), ETC

(Macedo et al., 2010) and MOODLE

platforms.

The teacher can choose the discussion forums to

mine the messages.

AUTOMATICANALYSISOFASYNCHRONOUSDISCUSSIONS

5 EXPERIMENTS CARRIED

WITH MINERAFÓRUM

To validate results using MineraFórum, five

experiments were made. In each experiment, a

discussion forum was analyzed by both the software

and two teachers. The forum topics were distinct, as

well as school level, and course modality. Forums

selected for the experiments were extracted from

ROODA, ETC and MOODLE platforms.

The purpose of the experiments was to compare

the average of message relevance calculated by

MineraFórum with the average obtained in the

assessments made by teachers. It is worthy

observing that the program calculates a relevance

value between zero and five for each text

contribution. For that reason, teachers were

requested to use the same values in their evaluations.

Evaluation of the messages was made by the

following group of teachers: two Ph.Ds and four Ms

teachers. Two teachers have long experience in

distance learning, and four have little experience.

Tables 2 and 3 describes the characteristics of the

forums used in the experiment: the VLE in which

they were offered, the discussion theme, the course,

the school level, and the modality of each course.

Tables 4 and 5 present values obtained in the

analysis made by both MineraFórum and teachers.

Information in this table includes: the teacher who

assessed the messages, the number of students who

participated in each forum, the number of posts, the

average obtained from the software analysis, the

average obtained from the teachers’ appraisals, and

the degree of similarity between the averages. The

average of analysis of the MineraFórum and the

average of the teachers’ assessments were obtained

by summing the values assigned to each post,

dividing by the number of messages. The degree of

similarity was obtained dividing the two averages.

Table 2: VLE and theme of analyzed forums.

Forum VLE Theme

1 Rooda Learning as transformation

2 Rooda Learning as transformation

3 ETC Team work

4 ETC Competence Development

5 Moodle Digital Ceritification

Tables 4 and 5 present the degree of similarity

between the average of the analysis performed by

MineraFórum and the one obtained in the analysis

made by teachers. This result reveals that the

average of the analysis calculated by the software is

similar to the average of the assessments made by

teachers.

Table 3: Characteristics of analyzed forums.

Forum Course School Level Modality

Education

(group 1)

Undergraduate Distance Education

Education

(group 2)

Undergraduate Distance Education

3 Extension Extension Face to face education

4 Extension Extension Face to face education

Informatio

n Systems

Extension Face to face education

Table 4: Analysis by MineraFórum and by Teacher 1.

Forum 1 2 3 4 5

Teacher A A C C E

Number of students 28 31 18 11 12

Number of

messages

48 73 76 42 12

Average of

analysis by

MineraFórum

2,92 2,88 3,00 3,21 3,42

Average of

analysis by teacher

2,79 2,32 3,61 3,90 2,67

Degree of

similarity

between analyses

by MineraFórum

and teacher

95,71% 80,48% 83,21% 82,32% 78,05%

In Table 5, the degree of similarity in forum 3

presented the least value, 76,00%. Analysis of forum

2, as shown in Table 5, obtained the highest value,

96,77%.

Table 5: Analysis by MineraFórum and by Teacher 2.

Forum 1 2 3 4 5

Teacher

B B D D I

Number of

students

28 31 18 11 12

Number of

messages

48 73 76 42 12

Average of

analysis by

MineraFórum

2,92 2,88 3,00 3,21 3,42

Average of

analysis by teacher

3,29 2,97 3,95 4,12 2,92

Degree of

similarity

(analyses by

MineraFórum and

teacher)

88,61% 96,77% 76,00% 78,03% 85,37%

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

It is relevant to remind that the software uses

three criteria to analyze messages: thematic

relevance of the text contribution, relevance of

message reference, and similarity of the post. When

assessing the posts, each teacher used their own

criteria.

Figure 7 presents the variation of the degree of

similarity found in the analysis of the five discussion

forums. Graphic “Similarity1” corresponds to the

degree between the averages obtained from Teacher

1 and from MineraFórum. Graphic “Similarity2”

refers to the comparison with Teacher 2.

Figure 7: Variation in the degree of similarity in the

experiments.

It was found that in some situations, the values of

the relevance of each post calculated by

MineraFórum were different of more than 2 points

regarding to the analysis of teachers. This occurred

in the following situations:

a) The message cited relevant concepts to the topic,

but not had value in the debate.

These cases occurred in posts where there was no

coherence and cohesion in the typed text. The

MineraFórum not analyses these parameters. The

software calculated the relevance of messages

according to the important concepts mentioned. The

teacher assigned a low value for these posts.

b) The message mentioned relevant concepts, which

were not cited in the reference text or in the forum.

In this situation, the teacher assigned a high value

for the post. As there were no conditions for the

software to identify these concepts, the calculated

relevance was low.

c) The post did not mentioned relevant concepts to

discussion, but had an important example of a

personal experience.

The tool assigned low importance to these messages.

The teacher analysed these posts with high

relevance.

d) The post did not cited relevant concepts to

discussion, but indicated an important bibliographic

reference or site.

The MineraFórum not analyzes the importance of

references or sites. Thus, the software calculated low

relevance to the post and the teacher indicated a high

value.

e) The message did not mentioned relevant

concepts to discussion, but attached an important file

or image.

The MineraFórum not analyzes the content of files

or images. Thus, the software calculated low

relevance to the post and the teacher indicated a high

value.

6 CONCLUSIONS

MineraFórum is a resource aimed at helping teachers

in qualitative analyses of forum posts. The software

performs the automatic execution of the

aforementioned activities. In such situations, the

teacher’s role in the debates is seen as crucial.

Considering the results presented in section 5,

the objective of the experiments was reached. It was

possible to verify that the average of the analysis of

the messages calculated by the software is similar to

the average obtained with appraisals made by

teachers.

MineraFórum is able to present the teacher with

a picture of the contributions made by learners, by

organizing and clustering the posts of the students.

This means that the program can provide

information that may be helpful to teachers in their

tutorial activities.

With information given by MineraFórum, the

teacher can guide his/her support to students who

posted few relevant contributions in a forum.

Teachers can also stimulate interactions between

learners who posted more relevant messages and

those who contributed with few relevant ones.

REFERENCES

Azevedo, B. F. T., Behar, P. A., Reategui, E. B., 2011a.

Automatic analysis of messages in discussion forums.

In Proceedings of the 14th International Conference

on Interactive Collaborative Learning. Pieštany,

IEEE, pp. 76-81.

Azevedo, B. F. T., Behar, P. A., Reategui, E. B., 2011b.

Um software para análise das mensagens de fóruns de

discussão. In Proceedings of the IADIS Ibero-

Americana WWW/Internet (CIAWI) Conference. Rio

de Janeiro, IADIS Press, pp. 195-202.

AUTOMATICANALYSISOFASYNCHRONOUSDISCUSSIONS

Behar, P. A., Bernardi, M., Souza, A. P. F. de Castro,

Kellen, K., 2007. ROODA: desenvolvimento,

implementação e validação de um AVA para UFRGS.

In XII Taller Internacional de Software Educativo.

Santiago, Chile, LOM Ediciones S.A., v. 1, pp. 321-

338.

Feldman, R., Sanger, J., 2007. The Text Mining

Handbook: Advanced Approaches in Analyzing

Unstructured Data, Cambridge University Press.

Garrison, D. R., Anderson, T., Archer, W., 2000. Critical

inquiry in a text-based environment: computer

conferencing in higher education. The Internet and

Higher Education, 2(2-3), pp. 87-105.

Gilbert, P. K., Dabbagh, N., 2005. How to structure online

discussions for meaningful discourse: a case study.

British Journal of Educational Technology, v. 36, n. 1,

pp. 5-18.

Li, Y., Dong, M., Huang, R., 2008. Semantic Organization

of Online Discussion Transcripts for Active

Collaborative Learning. In Proceedings of the IEEE

International Conference on Advanced Learning

Technologies. IEEE Computer Society Press, pp. 756-

760.

Macedo, A. L. et al., 2010. ETC: o que mudou e por quê?.

Revista Novas Tecnologias na Educação, v. 8, pp. 2-

12.

Palloff, R. M., Pratt, K., 2002. Construindo comunidades

de aprendizagem no ciberespaço, Artmed. Porto

Alegre.

Palloff, R. M., Pratt, K., 2004. O aluno virtual: um guia

para trabalhar com estudantes on-line, Artmed. Porto

Alegre.

Puretskiy, A. A., Shutt, G. L., Berry, M. W., 2010. Survey

of text visualization techniques. In Berry, M. W.,

Kogan, J. Text mining: applications and theory. John

Wiley & Sons Ltd, pp. 107-127.

Ravi, S., Kim, J., 2007. Profiling Student Interactions in

Threaded Discussions with Speech Act Classifiers. In

Proceedings of the AI in Education Conference

(AIED), Los Angeles.

Rebedea, T., Trausan-Matu, S., Chiru, C., 2008.

Extraction of Socio-semantic Data from Chat

Conversations in Collaborative Learning

Communities. Dillenbourg, P., Specht, M. (eds.). EC-

TEL 2008. LNCS, Springer-Verlag Berlin Heidelberg,

n. 5192, pp. 366-377.

Tan, A., 1999. Text Mining: The State of the Art and the

Challenges. In Proceedings of the PAKDD’99

Workshop on Knowledge Discovery from Advanced

Databases, pp.71-76.

Torre, C. J. de la, Martín-Bautista, M. J., Sánchez, D.,

Vila, M. A., 2005. Text mining: Intermediate forms

for knowledge representation. In Proceedings of the

Joint Eusflat/Lfa Conference. Montseny, E.,

Sobrevilla, P. (eds.), European Society for Fuzzy

Logic and Technology, pp. 1082-1087.

MineraFórum is part of the research projects:

“MineraROODA: Ferramentas de mineração de conteúdo

cognitivo e de subjetividade afetiva no Ambiente Virtual

de Aprendizagem ROODA”, Announcement MCT/CNPq

014/2010 - Universal; “ROODA: novas ferramentas para

incorporação no ambiente virtual de aprendizagem”,

Researcher Gaúcho Program, Announcement FAPERGS

006/2010. “Ampliando possibilidades pedagógicas através

da tecnologia de mineração de textos integrada à escrita

coletiva a distância”, Announcement MEC/CAPES

029/2010.

WordNet is a large lexical database of English; available

at: http://wordnet.princeton.edu

iii

Núcleo de Tecnologia Digital aplicada à Educação / Pós-

graduação em Informática na Educação.

ROODA is one of the platforms used for Distance

Education at UFRGS. ROODA is available at:

https://www.ead.ufrgs.br/rooda/

ETC is a collective text editor develped by NUTED, and

used at UFRGS as well as in extension courses, available

at http://www.nuted.ufrgs.br/etc2/

MOODLE, with MineraFórum integrated into it, is

available at http://www.nie.iff.edu.br/moodle/

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation