LLM-Assisted Thematic Analysis Protocol (LATA)
Phase 2 — Concept Integration (v1.0)

1. Purpose

This phase aims to integrate the latent topics identified during Concept Discovery into a coherent set of higher-order macroclusters.

Rather than discovering new research phenomena, the objective is to identify broader conceptual domains that emerge from the relationships among latent topics. Macroclusters provide an intermediate level of thematic abstraction that preserves analytical richness while improving the interpretability of the overall thematic structure.

2. Input

The input consists of the validated inventory of latent topics produced during Phase 1 — Concept Discovery.

Each latent topic should include:

Topic label
Conceptual description
Representative concepts
Representative publications
Notes on conceptual boundaries or ambiguities

The analysis should be performed at the latent-topic level. Individual bibliographic documents should only be revisited when necessary to clarify uncertain conceptual relationships.

3. Analytical Principles

This phase follows five methodological principles.

a) Conceptual affinity. Macroclusters should be constructed according to conceptual affinity rather than lexical similarity. Topics may differ substantially in terminology while representing complementary aspects of the same broader research domain.

b) Preservation of analytical granularity. Integration should preserve the analytical identity of each latent topic. Macroclusters organize latent topics but do not replace them.

c) Conceptual coherence. Each macrocluster should represent a well-defined conceptual domain with strong internal consistency and clear conceptual boundaries.

d) Balanced abstraction. Macroclusters should simplify thematic interpretation without oversimplifying the diversity of the underlying latent topics.

e) Interpretative transparency. Where more than one plausible thematic organization exists, the rationale supporting the selected organization should be explicitly documented.

4. Operational Instructions

Conduct a higher-order conceptual integration of the latent topics identified during Concept Discovery.

The analysis should proceed through four sequential stages.

Stage 1 — Global Review

Review the complete inventory of latent topics to develop a comprehensive understanding of the thematic landscape before proposing any thematic integration.

Focus on conceptual relationships among topics rather than on individual topic descriptions.

Stage 2 — Evaluation of Conceptual Affinities

Identify latent topics that exhibit strong conceptual affinity.

Evaluate affinity according to:

shared research phenomena;
complementary analytical perspectives;
common policy domains;
related societal challenges;
overlapping research questions.

Do not rely on lexical similarity alone.

Stage 3 — Macrocluster Construction

Construct macroclusters by integrating latent topics exhibiting strong conceptual affinity.

Each macrocluster should:

represent a recognizable research domain;
maintain internal conceptual coherence;
preserve meaningful distinctions among constituent latent topics;
remain clearly distinguishable from other macroclusters.

Assign each latent topic to the macrocluster that best represents its primary conceptual contribution.

Stage 4 — Internal Critical Review

Critically evaluate the proposed macrocluster structure.

Review:

conceptual coherence within each macrocluster;
conceptual separation between macroclusters;
latent topics with potentially ambiguous assignments;
macroclusters that appear excessively broad or internally heterogeneous.

Where necessary, revise the thematic organization while preserving analytical consistency and interpretability.

5. Expected Output

The output of this phase should include:

a structured inventory of macroclusters;
a descriptive label for each macrocluster;
a conceptual definition of each macrocluster;
the latent topics assigned to each macrocluster;
a rationale explaining the conceptual affinity supporting each grouping;
notes regarding conceptually ambiguous assignments, where appropriate.

The objective is to produce an interpretable thematic organization that serves as the conceptual foundation for higher-level abstraction.

6. Quality Assurance Checks

6.1. Analytical calibration

Before accepting the macrocluster structure, manually review a conceptually representative sample of latent topics assigned to each macrocluster. Pay particular attention to latent topics whose assignment could plausibly belong to more than one macrocluster or whose conceptual boundaries appear uncertain.

Assess whether each grouping reflects genuine conceptual affinity rather than superficial thematic similarity. Where systematic inconsistencies are detected, document representative cases, explain the conceptual rationale supporting an alternative organization, refine the integration criteria, and repeat the clustering process until the resulting macroclusters exhibit satisfactory conceptual coherence and internal consistency.

6.2. Phase-specific checks

Before completing this phase, verify that:

every latent topic has been assigned to a macrocluster;
assignments are supported by conceptual affinity rather than lexical similarity;
macroclusters remain internally coherent and externally distinguishable;
the analytical identity of latent topics has been preserved;
conceptual uncertainties have been documented where appropriate;
the resulting macroclusters provide an appropriate intermediate level of abstraction for subsequent conceptual synthesis.

7. Transition to Phase 3

The validated inventory of macroclusters constitutes the input for Phase 3 — Concept Abstraction.

The objective of the next phase is to identify higher-order conceptual dimensions that organize the macroclusters into a coherent representation of the overall intellectual structure of the bibliographic corpus.