Context and method

Four different senior high schools (SHSs) from Ghana participated in the study.

Three of the schools were intermittently involved in the development of the EmoTect system. The schools are Mfatsipim SHS, Osei Kyeretwie SHS, Bompata SHS and Agona Seventh Day Adventist SHS. The participants from the schools were involved in the design of the EmoTech system as part of participatory design (PD) approach taken. PD is a design approach that attempts to actively involve stakeholders (thus, students and counsellors) in the design process, thereby ensuring that the results based on the design meets stakeholder’s expectation. According to Roberston and Simonson (2012: p.6), “during a Participatory Design process all participants increase their knowledge and understandings: about technology for the users, about users and their practice for designers, and all participants learn more about technology design”. Duveskog et al., (2009) undertook a similar project in Tanzania regarding a digital tool for counselling people living with HIV/AIDS. Duveskog et al. (2009) adopted a PD consisting of teams of secondary school children, university counsel-lors, HIV counselling experts and experts in ICT were involved in the implementa-tion process.

Counsellors and randomly selected students from the various schools partici-pated in the project intermittently. Apart from the main counsellors, auxiliary coun-sellors from the various schools were also involved at some stages. A number of se-lected teachers participated in the project as well, especially during the contextual evaluation of the project. In all the varying stages of the project, where necessary, the participants were contacted in their various schools for their input. Questionnaires, interviews and observations were the main research techniques used in the collection of data throughout the studies. Preliminary data collections from the participants were mostly published in [PI] to [PIV]. The outcome influenced the development of EmoTect which ultimately led to [PV].

As indicated in Table 3.1, RQ1 (What are the emotional challenges that threatens stu-dents’ academic pursuit?), RQ2 (What counselling technologies are being used in the senior high schools of Ghana?), RQ3 (What are the factors that motivate students to adopt and use e-counselling in Ghana?) and RQ4 (Does the emotional state of counsellors influence their emotion perception while annotating emotions in text?) form part of the questions that led to understanding of the research environment. RQ1 and RQ2 were qualitatively in-vestigated and the technique for data collection was through the use of question-naires and interviews. In RQ3, we adopted a quantitative research method where statistical research approach was used. The study was investigated in [PII] where

68

only questionnaires were used to collect data from the selected students. Mixed re-search method was used in [PIV] where RQ4 (Does the emotional state of counsellors influence their emotion perception while annotating emotions in text?) was investigated.

With this, only questionnaires were used in data collection. RQ5 (How can a text-based emotion and sentiment classification system be constructed for counselling delivery?) and RQ6 (How can a text-based emotion and sentiment classification system for counselling be evaluated?) were both investigated in [PV]. While RQ5 was in line with constructiv-ism approach of requirement elicitation and implementation of the requirements in EmoTect system, RQ6 was the evaluation of the EmoTect system. The technique for gathering data in the RQ5 and RQ6, as seen on the table, are through interview, ques-tionnaires and observation.

Table 3. 1. Research methods and techniques for each research question

3.2.1 Counsellors’ profiles

Three of the counsellors from the three active schools who participated in the study have ages ranging from 30 to 50 years. Two of them have bachelor’s degrees in edu-cational counselling, while the other one has a bachelor’s degree in history. The one with the history background was once a teacher and subsequently was converted into a school counsellor because the sector lacks professional counsellors to help stu-dents. All the main counsellors reside on their respective school campuses where they counsel students and organise other symposia for students and teachers. The counsellors have worked in their respective schools for over 4 years as counsellors.

Two of the counsellors are male, while the other one is female.

3.2.2 Requirement elicitation

Coupled with our preliminary findings, we organised a semi-structured interactive discussion session with the selected counsellors and students before the commence-ment of EmoTect’s developcommence-ment, forming part of the DSR requirecommence-ment. One counsel-lor and students, each from the three selected schools in Ghana, participated in the discussion. The session was organised in the respective schools of the participants.

Altogether, three counsellors and thirty students contributed to the discussion. The essence of the meeting was to outline to the participants the preliminary

require-Research Method DSR process Research technique

RQ1 Qualitative research

Analysis and re-quirement elicita-tion

Questionnaire and interview

RQ2 Qualitative research Questionnaire

RQ3 Quantitative research Questionnaire

RQ4 Mixed research Questionnaire

RQ5 Constructivism/design Implementation Interview, observation RQ6 Mixed research Evaluation Observation, questionnaire

69 ments for the commencement of EmoTect’s development based on previous encoun-ters. By so doing, this researcher expected the participants to contribute to the dis-cussion before taking the next step to implement the initial (pre-defined) require-ments for EmoTect. The initial preliminary requirerequire-ments and brief description of the various components of EmoTect, as discussed with the participants, are shown in Table 3.2.

Table 3.2. Requirements from researcher’s preliminary studies

Requirement Description

Contact counsellor’s widget form Presentation layer where students can contact counsellors on the grounds of seeking counselling (anonymous).

Emotion extraction and visualisation The domain logic that extracts and visualises emo-tions of students’ submissions automatically.

Sentiment extraction and visualisation Where emotional valence (sentiments) from stu-dents’ submissions are to be extracted. Only posi-tive and negaposi-tive valence is considered.

While the aforementioned proposed requirements were welcomed, the partici-pants suggested further ideas that were worth considering in the development of EmoTect. Counsellors, for their part, requested that the intended system be able to assist them in monitoring the emotional changes of their students over time. This generated the idea of creating a database to store the extracted emotions’ categories (results) for future reference. Therefore, in designing the database, this author con-sidered two different functionalities of EmoTect.

On the one hand, the database is meant to house the annotated corpus (life stories of students) meant to be used for classifier training. On the other hand, the database is to store the extracted emotions for future reference, which is made available in visual form at the EmoTect interface, as proposed by the participants. Additionally, counsellors were concerned about emotion keywords from students’ textual submis-sions, which according to the participants, would prompt further review of students’

submissions should the need arise. This resulted in the idea of extracting and output-ting emotional keywords for counsellors at the EmoTect interface for users. The rea-son connected to the output of the emotional keywords is that counsellors will be able to consider students’ textual submission holistically, should there be any suspi-cious keywords. For instance, emotion keywords like kill, suicide, worry and die found in students’ submissions will give a clue to a counsellor about what a student may be up to. With the aforementioned emotion keywords, one may interpret that the student is either threatening suicide or he or she is only talking about suicide and death. Students expressed their concerns about the poor internet connection in the country, which partly forms the challenges associated with the implementation of

e-70

counselling in Ghana. This concern was raised given that EmoTect is a web-based platform.

Though the counsellors are not technically inclined, this researcher still explained to them the basic idea of training a classifier. This researcher has explained the con-cept as part of the requirement that users will be made to tag the training data based on their understanding of each instance of the training corpus. As expected, there were no opposing questions or ideas to that effect. The final requirements for the implementation were itemised for the development to commerce. In Chapter 4, we delve into the implementation of EmoTect based on the elicited requirements from the counsellors.