README — LATA Protocol User Guide (v1.0)

Overview

The LLM-Assisted Thematic Analysis (LATA) Protocol is a methodological framework that formalizes the conceptual reasoning underlying LLM-assisted thematic analysis of bibliographic corpora through a transparent, traceable, hierarchical, and expert-supervised analytical process.

Rather than documenting isolated prompts or model-specific interactions, LATA defines a structured protocol in which bibliographic documents are progressively transformed into an interpretable conceptual framework through successive stages of conceptual reasoning, thematic integration, abstraction, and validation.

The protocol and its methodological principles are intentionally model-agnostic, domain-independent, and reusable, allowing its application to thematic analyses across different scientific disciplines and bibliographic corpora.

1. Why LATA?

Large Language Models (LLMs) have rapidly become valuable analytical assistants for qualitative and thematic analysis. Nevertheless, many published studies describe only the prompts employed or provide general descriptions of the analytical workflow, offering limited information about the conceptual reasoning underlying the analysis.

As a consequence, reproducing, auditing, comparing, or extending these studies remains difficult. LATA addresses this methodological gap by documenting the reasoning process itself.

Rather than asking:

Which prompt generated these results?

LATA asks:

How should thematic reasoning be organized so that the resulting conceptual framework is transparent, traceable, reusable, and scientifically defensible?

The protocol therefore shifts attention from prompt engineering toward methodological design.

2. Protocol Design Principles

LATA is founded upon five foundational principles, which address how the Protocol is designed to guarantee quality, reusability, and robustness.

a) Transparency: Every analytical decision, prompt, intermediate output, and conceptual transformation is explicitly documented throughout the workflow, allowing users to understand how the final thematic framework has been constructed.

b) Traceability: The protocol preserves clear links between each analytical stage, enabling the progression from the original bibliographic corpus to topics, macroclusters, and conceptual dimensions to be reconstructed and independently reviewed.

c) Hierarchical organization: Conceptual knowledge is synthesized progressively through successive levels of abstraction, transforming fine-grained thematic evidence into increasingly parsimonious and interpretable conceptual structures.

d) Modularity: Each phase constitutes an autonomous analytical module with standardized inputs, outputs, quality assurance procedures, and transition criteria. This modular architecture facilitates adaptation, extension, substitution of individual phases, and interoperability with different analytical techniques or LLMs.

e) Expert-supervised AI: Large Language Models assist the analytical process by supporting interpretation and synthesis, while all conceptual decisions remain subject to expert review and validation to ensure methodological rigor and conceptual coherence.

3. Methodological Principles

LATA is founded upon five methodological principles, which address how the thematic analysis should be conducted.

a) Corpus-driven reasoning:

Conceptual structures emerge inductively from the bibliographic corpus rather than being imposed by predefined theoretical frameworks.

b) Hierarchical conceptualization:

Conceptual abstraction progresses through successive analytical levels, each answering a different scientific question while preserving analytical continuity.

c) Traceable transformations:

Every analytical output constitutes the documented input for the subsequent phase, allowing the complete conceptual reasoning process to remain auditable.

d) LLM-assisted, expert-supervised reasoning:

LLMs perform structured conceptual reasoning, whereas domain experts critically evaluate, challenge, refine, and validate the resulting conceptual framework.

e) Transparent methodological documentation:

Every analytical transformation is explicitly documented to maximize transparency, reproducibility, methodological reuse, and scientific accountability.

4. Repository Structure

| File                       | Purpose                                               |
| ---------------------------| ----------------------------------------------------- |
| README.txt                 | User Guide and methodological overview                |
| P1_Concept_Discovery.txt   | Bibliographic Documents → Latent Topics               |
| P2_Concept_Integration.txt | Latent Topics → Macroclusters                         |
| P3_Concept_Abstraction.txt | Macroclusters → Conceptual Dimensions                 |
| P4_Concept_Validation.txt  | Conceptual Framework → Validated Conceptual Framework |

Additional project-specific resources (datasets, scripts, figures, thematic outputs, and documentation) may accompany the protocol depending on the particular research application.

5. Using the Protocol

The protocol is designed for both human researchers and LLMs.

LATA should be executed sequentially through its four analytical phases.

Each phase receives the validated output produced by the previous phase and transforms it into a higher level of conceptual abstraction.

Each phase is provided as an independent document organized into seven sections. The section "Operational Instructions" contains the executable analytical prompt intended for submission to the selected LLM. The remaining sections provide the methodological context, analytical principles, expected outputs, quality assurance procedures, and transition guidelines required to correctly execute, interpret, and validate each phase. Although only the Operational Instructions need to be executed by the model, users are strongly encouraged to consult the complete document, as the additional sections provide essential methodological guidance for ensuring analytical consistency and expert supervision throughout the protocol.

However, the protocol is intentionally iterative rather than strictly linear.

Following the execution of each phase, the researcher should critically review the generated output before proceeding. Whenever substantial conceptual weaknesses, inconsistencies, or alternative interpretations are identified, the analytical rationale should be refined and, where necessary, the corresponding phase repeated before continuing with the workflow.

Consequently, LATA should be understood as a protocol for progressive conceptual refinement rather than a fixed sequence of prompt executions. Each phase combines LLM-assisted reasoning with expert critical review before the workflow proceeds to the next stage.

This way, the Protocol incorporates two mechanisms of iteration:

a) Micro-interation: within each phase

Task
↓
Analytical Calibration
↓
Refine instructions
↓
Repeat if necessary

b) Macro-iteration: among phases

Phase
↓
Expert Review
↓
Return to previous phase if required

6. The LATA Workflow

LATA can be viewed from two complementary perspectives.

a) User workflow

Prepare bibliographic corpus
↓
Execute Phase 1
↓
Expert review
↓
Execute Phase 2
↓
Expert review
↓
Execute Phase 3
↓
Expert review
↓
Execute Phase 4
↓
Scientific interpretation

b) Internal conceptual workflow:

Bibliographic documents
↓
Latent topics
↓
Macroclusters
↓
Conceptual dimensions
↓
Validated conceptual framework

Both perspectives describe the same analytical process: the first emphasizes the research workflow, whereas the second represents the conceptual transformations performed throughout the protocol.

7. Hierarchical Conceptualization

| Analytical Level        | Primary Output       | Scientific Question
| ------------------------| -------------------- | ----------------------------------------    
| Bibliographic documents | Corpus evidence      | What has been published?              
| Latent topics           | Research phenomena   | What is being studied?             
| Macroclusters           | Research domains     | Which research phenomena belong together?    
| Conceptual dimensions   | Conceptual framework | How has the literature collectively conceptualized the phenomenon?

Rather than representing a hierarchy of thematic aggregation, these analytical levels constitute a hierarchy of conceptual explanation, with each level contributing a distinct form of scientific understanding.

8. Relationship with the Research Workflow

The analytical workflow described in the associated scientific study and the LATA Protocol represent two complementary descriptions of the same research process.

The manuscript documents the statistical and computational procedures used to analyse the bibliographic corpus, whereas LATA formalizes the conceptual reasoning underlying thematic interpretation.

Consequently, the protocol complements rather than replaces the methodological workflow reported in the associated publication.

| Research workflow                                      | LATA Protocol                 |
| ------------------------------------------------------ | ----------------------------- |
| Corpus characterization and exploratory analysis       | Phase 1 — Concept Discovery   |
| Topic modelling                                        | Phase 2 — Concept Integration |
| Construction of macroclusters and conceptual synthesis | Phase 3 — Concept Abstraction |
| Expert validation and methodological review            | Phase 4 — Concept Validation  |

9. Validation Philosophy

The objective of LATA is not to demonstrate that a thematic framework is uniquely correct. Instead, the protocol seeks to produce a conceptual framework that is:

Internal quality
Conceptually coherent
Conceptually robust
Scientific usability
Transparent
Interpretable
Fully traceable

These properties are reinforced through two transversal principles:

a) Reflexivity, encouraging continuous critical reflection on analytical decisions.

b) Methodological defensibility, ensuring that every conceptual transformation remains scientifically justified and explicitly documented.

Expert validation is therefore conceived as an expert-challenged critical review, whose purpose is to critically evaluate and strengthen the conceptual framework rather than merely confirm the original LLM output.

10. Scope and Applications

LATA is intended for thematic analyses of large bibliographic corpora, including:

Bibliometric studies
Systematic literature reviews
Scoping reviews
Evidence syntheses
Large-scale qualitative analyses of scientific literature

Although originally developed using GPT-5.5, the protocol is independent of any particular LLM implementation and may be adapted to future language models or equivalent AI systems capable of supporting structured qualitative reasoning.

11. Limitations

LATA does not eliminate the interpretative nature of thematic analysis.

Alternative conceptual organizations may remain plausible, both across researchers and across different executions using LLMs.

This characteristic is not unique to LLM-assisted analysis but reflects the interpretative nature of qualitative thematic research more generally.

The protocol therefore seeks not to eliminate analytical uncertainty, but to make it transparent, traceable, critically reviewed, and methodologically defensible through explicit documentation and expert supervision.

12. Citation

When using the LATA Protocol, please cite as follows:

Orduña-Malea, E. (2026). LATA Protocol: A Methodological Framework for LLM-Assisted Thematic Analysis of Bibliographic Corpora. Available at:
https://riunet.upv.es/handle/10251/236921