Section A

Question 1

Guidelines for this End-of-Course Assessment report are as follows:

1. Draft a report on your proposed topic focusing on the practical application of text mining in a specific field or industry. The materials and sources used should substantially surpass the content covered in this course or any other ANL courses.

2. Construct a text mining project based on your selected topic.

3. You can choose either one of the following two options:

•  Option 1: Apply text categorisation using the IBM SPSS Modeler. You will need to find a dataset with minimum 100 rows of text records for this option. Your response should include documentations of the effort put into improving the resource template and creating the categories from scratch (refer to the GBA steps). Provide screenshots of the text (5-8 samples) for each category, showing your effort to correctly categorise the text. Do not use the ‘Build Categories’ feature that automatically builds the categories as no credit will be given if this is used. Similarly, do not use the “Text Analysis Package”.
• Option 2: Apply topic modelling using R programming. You have the option to include sentiment analysis, but this is not mandatory. You will need to find a dataset with minimum 500 rows of text records for this option. Using other software, e.g., IBM SPSS Modeler, for necessary data preparation is allowed for this option.

For topic modelling, please ensure the following:

• Customize the stop words.
• Try a few different values of k and document the process in your answer.
• Evaluate the appropriateness of your final topic model. Include screenshots of the top 5 texts for each topic discovered; present them in your report, not in the appendix. To do this, sort the gamma values of each topic in the gammaDF table, and refer to the text in your original dataset using the doc id.
• Name each topic appropriately.
• More credits will be given if you improve the model by incorporating additional steps beyond the example given in the Study Guide.

• Describe the R package you selected for this project, especially if it differs from the one used in the Study Guide.
• Evaluate the appropriateness of the sentiment scores generated by the selected R package based on the text you collected.
• Include appropriate screenshots of the texts after sentiment analysis along with their corresponding sentiment scores/polarity. Present the screenshots with the first 30 rows in your report, not in the appendix.
• Please note that sentiment analysis is optional. Including it may or may not earn additional credits, depending on the quality of the analysis.

4. Possible sources of references for your report:

•  Internet websites

http://www.kdnuggets.com/index.html

http://www.dextra.sg/

https://github.com/stepthom/text_mining_resources

• Journal articles (Use SUSS library (https://library.suss.edu.sg/) or
  Google Scholar).
• Conference papers especially those from the SAS Global Forum where
  they feature text mining applications

https://www.sas.com/en_us/events/sas-global forum/program/proceedings.html

• Please write between 3,000 to 5,000 words (excluding cover page, table of content, reference and appendices). Marks will be deducted for those that are below 3000 words. For those reports that exceed 5000 words, only the part that is below 5000 words will be graded and the rest will be ignored. No word limit for individual section.
• Font size 12, Times New Roman, 1.5 lines spacing.
• Reference citation and reference list: Use American Psychological Association (APA) referencing style. Please refer to ANL312 Study Unit 1 for details.
• You must acknowledge and reference all sources used.

Topic Selection

1. You are required to select a topic of your choice. A list of topics is provided in Appendix A. You may propose a topic that is not on the list. However, your topic must be related to things you have learnt in this course.

2. Once you have selected the topic, conduct research to look for a dataset for your project and references for your Literature Review section.

3. Please note that any “hotel review” data is not allowed for this year’s ECA.

Question 1a

Introduction: Based on your chosen topic, discuss the project background, the business analytics concepts/issue(s) involved, the project objective(s) and the role of text mining in achieving the objective(s). Define all relevant terms used (e.g., text mining, N-grams).

Question 1b

Literature Review: Describe two (2) references (must be research articles from journal/conference/academic report/thesis) that apply text mining in a way relevant to your selected topic. Include the general background of the study, the dataset used, the details of the text mining process applied, as well as relevant findings and conclusions. Discuss the implications of the references to the current project.

Question 1c

Body: Use the CRISP-DM framework to organize your report. You are required to find a small dataset and construct a text mining model to achieve your project objective(s).

Question 1d

The last two sections of the report include:
Summary: Summarise the key findings, insights, and conclusions obtained from your text mining analysis.
References: List all the sources you cited in your report and follow the APA
referencing style.

Additionally, the entire report will be evaluated based on the coherence and balance maintained across all sections. The Introduction should provide a clear motivation for the project. The Literature Review section should thoroughly review materials closely related to the project. In the CRISP-DM section, the steps should be logically presented, demonstrating a sound approach to text mining. The Summary section should effectively wrap up the project, highlighting key findings and insights. The references should be relevant and support the key ideas presented. The writing should be professional, with good and plain English, and adhere to all the instructions given.