Version 2: all 201 supplied source pages have been compared individually. Transcription corrections and unresolved original source issues are documented separately.
Reading and citing the thesis today
Editorial companion, 30 October 2026. This guide is separate from the 2003 thesis.
Tanat Tonguthaisri’s Enhancing Academic Advisement with Case-Based Reasoning describes a prototype that recommends courses using previous cases and applies rules to course requirements. The thesis is useful to readers studying how an advisory system represents experience, ranks similar situations, and handles constraints. Its conclusion also addresses the work needed to maintain a case base and the limits of replacing human advice. Those are specific reasons to read and cite the work; this edition makes no claim of priority over all other academic-advising research.
Start with the question you are studying
| Research question | Relevant thesis material | Useful interpretation |
|---|---|---|
| How can prior advising experience inform a new recommendation? | Chapter 2, especially §§2.10.4–2.10.6; Chapter 4 prototype | Examine case representation, similarity and retrieval. Cite the historical method actually described. |
| How should hard course requirements interact with preferences? | Chapter 5, §5.1, printed p140 (PDF p149) | The thesis assigns core-course and prerequisite filtering to rules, and elective advice to CBR. This distinction supports discussion of constrained recommendation. |
| What makes an advising system maintainable? | §4.8.2.2.2, printed p134 (PDF p144); §§5.2.1–5.2.2 | Read the discussion of advisor participation, continued knowledge acquisition, case representation and case-base growth. These are design considerations, not measured deployment benefits. |
| How can a learner refine an initial recommendation? | §§4.8.3–4.8.5, printed pp136–138 (PDF pp145–147) | Tweaking, ratings and preference feedback provide historical examples for interactive recommender design. |
| What did the prototype establish? | Chapter 4 evaluation and Chapter 5, printed pp139–142 (PDF pp148–151) | Read the evaluation with its assumptions. The conclusion describes a small proof of concept, clear-pass students and a single program stream. Stated advantages should not be converted into measured effect sizes. |
The source conclusion explicitly calls for academic-advisor crosschecks and notes exceptional cases. A new interactive demonstration on this website uses invented examples. Its output is neither educational advice nor a replication of the thesis evaluation.
Connections to later work
CBR provides a vocabulary for reasoning from previous cases, adapting solutions and retaining experience. Aamodt and Plaza’s 1994 paper is the appropriate starting point for the general framework. Watson and Marir’s review discusses representation, retrieval, adaptation and practical systems. These sources explain the framework used in the thesis, rather than assigning the thesis authorship of CBR itself. Aamodt and Plaza; Watson and Marir.
A comparison with retrieval-augmented generation can help current AI readers, provided the distinction is explicit. Lewis and colleagues’ 2020 RAG paper combines a neural language generator with retrieved external passages. The thesis retrieves structured cases to support course recommendations. Both involve retrieval; their representations, adaptation mechanisms and output generation differ. This comparison is a 2026 interpretation, not a claim that the 2003 thesis implemented or predicted LLM-based RAG. Lewis et al., 2020.
For a recent application comparison, Ma, Hahsler and Moore’s 2025 curriculum-advising paper describes an LLM-based recommender architecture involving curriculum requirements, prerequisite dependencies and student preferences. It offers a later design to compare with the thesis’s rules-and-cases approach. This reading list does not establish that the later authors cited, used or were influenced by this thesis. Ma et al., 2025.
Citation and edition identity
Use the repository citation for the historical research:
Tonguthaisri, T. (2003). Enhancing Academic Advisement with Case-Based Reasoning [Master of Engineering thesis, RMIT University]. https://doi.org/10.25439/rmt.34846194
The thesis title page says August 2003. The RMIT record became public on 30 October 2026. The original DOI identifies the repository thesis; it does not imply that RMIT reviewed or endorsed this website, the reconstructed diagram, the transcription, or the synthetic example. When quoting this edition, check the scanned source and cite the printed page, with the physical PDF page if clarification is needed. For new diagrams or website explanations, also identify the 2026 reconstruction and its URL. RMIT repository record.
The supplied scan has 201 physical PDF pages. Its printed numbering jumps from 134 to 136 between physical pages 144 and 145. The next section heading is consecutive. Whether this is original mispagination or an omitted page is unresolved; the reconstruction does not supply invented content. Editable text remains a transcription with recorded corrections and source links, not a newly certified critical edition.
Discovery that supports accurate citation
The website provides a dedicated thesis landing page, a complete static HTML reader, direct PDF access and citation downloads. Scholarly metadata preserves the title, author, institution, 2003 publication year and original DOI. Repository-entry dates are separate. Google Scholar’s guidelines call for searchable PDF or HTML, accessible abstracts and recognizable bibliographic metadata; they also specify a 5 MB file limit. These are implementation criteria, not evidence that this site is already indexed. Google Scholar inclusion guidelines.
Use natural vocabulary across the explanatory pages: case-based reasoning, academic advising and advisement, educational recommender systems, knowledge representation, decision support, similarity retrieval and case-base maintenance. The repository already records the three FoR categories 460206, 460902 and 460105. Avoid adding modern terms to the original thesis title or abstract. General search visibility depends on useful content, understandable links and successful crawling; a sitemap or structured data cannot guarantee placement. Google Search SEO guide.
For follow-through, link the exact DOI from the author’s scholarly profiles and personal publication list. ORCID supports adding works through Search & Link services and importing BibTeX. Check title, author, institution and year after import. ORCID guidance.
If another repository is used later, distinguish a copy of the same work from a new supplement. Zenodo asks for the existing DOI when the same object already has one, and warns that a supplementary dataset should not reuse the article’s DOI. A separately authored dataset, software package or retrospective essay should identify its relation to the thesis. Do not apply the blanket rule “never create another DOI” to every derivative object. Zenodo DOI guidance.
Search-engine marketing can bring visitors, but visits and ad clicks are not scholarly citations. No advertising expenditure or profile submission is part of this reconstruction. Review indexing after publication, then track downloads, relevant search queries and citations separately. Establish a dated baseline before claiming improvement; no citation baseline or forecast was established for this edition.
Rights and reuse
The repository records © Tanat Tonguthaisri 2003 and “All rights reserved.” Public access is not a CC-BY grant. This author-directed reconstruction preserves that status; it does not relicense third-party material. The supplied SVG is an author-provided 2026 explanatory reconstruction with Watson extensions, replacing a redacted figure. Its placement and provenance are disclosed. The original redacted repository PDF remains available for comparison.
Machine-readable formats
The package supports human reading through PDF and EPUB; editing through Word and LaTeX; and machine use through Markdown, page-anchored JSON/JSONL, structured XML and bibliographic exports. Preserve original source text, corrected text and editorial notes in distinct fields. Use physical PDF pages as stable anchors even when printed numbering is uncertain. XML structure helps downstream processing but does not make OCR correct or automatically place a thesis in an index.
When building a retrieval system from these files, return source pages with answers, keep historical code as archival text unless independently repaired and tested, and distinguish synthetic examples from the thesis’s evidence. The accompanying TypeSafe guide demonstrates semantic image checks and their failure cases. Exact formulas, page maps, source hashes and citations remain deterministic checks.
Download this guide as Markdown · Explore TypeSafe image QA
Download selected reading in BibTeX. This is the 2026 companion reading list, separate from the thesis’s original bibliography.