LN: Kitchenham & Charters (2007) — Guidelines for Systematic Literature Reviews in SE

Bibliographic Reference

Citation: Kitchenham, B., & Charters, S. (2007). Guidelines for performing systematic literature reviews in software engineering (Technical Report EBSE-2007-01). Keele University.


Pass 1 — Bird’s Eye View (5 Cs)

CAssessment
CategoryMethodological guidelines (technical report)
ContextEvidence-Based Software Engineering group at Keele University. Adapts Cochrane Collaboration’s systematic review protocol for the SE domain
CorrectnessThe most widely cited SLR guideline in software engineering (8,000+ citations); basis for PRISMA-adapted SE reviews
Contributions(1) SLR protocol with three phases (Planning, Conducting, Reporting); (2) Research question formulation (PICO framework); (3) Search string construction; (4) Inclusion/exclusion criteria; (5) Data extraction forms; (6) Quality assessment checklist
ClarityVery good — structured, prescriptive, with worked examples

Relevance: ⭐⭐⭐⭐⭐

PUMA’s SLR (literature review phase) follows Kitchenham & Charters (2007) as its methodological basis. The search protocol, inclusion/exclusion criteria, and PRISMA logging in PUMA’s PRISMA-Log directly implement this guideline.


Pass 2 — Key Concepts

The Three SLR Phases

Phase 1: Planning

  • Define the need for a review (is an SLR necessary or does one exist?)
  • Define research questions using PICO: Population, Intervention, Comparison, Outcome
  • Develop review protocol (document before execution — the key rigour mechanism)

PUMA’s Research Questions (PICO format):

ElementPUMA SLR
PopulationLLM-based agents applied to PM tasks
InterventionPrompting strategies (Zero-Shot, Few-Shot, CoT)
ComparisonHuman baseline + rule-based systems
OutcomeF1-macro (triage), MAE (estimation), statistical significance

Phase 2: Conducting

  1. Search string construction: Combine keywords with Boolean operators across multiple databases
  2. Source selection: IEEE Xplore, ACM DL, arXiv, Semantic Scholar
  3. Study selection: Title → Abstract → Full-text screening against inclusion/exclusion criteria
  4. Data extraction: Structured form per included study
  5. Quality assessment: Internal validity, construct validity, external validity

PUMA’s inclusion criteria:

  • Published 2019–2025
  • LLMs applied to software engineering or PM tasks
  • Empirical evaluation with measurable outcomes
  • English language

PUMA’s exclusion criteria:

  • Pure theoretical/survey papers without empirical component
  • Applications outside software/PM domain
  • No reproducible methodology

Phase 3: Reporting

  • Narrative synthesis of findings
  • PRISMA flow diagram (search → screened → included → analysed)
  • Evidence tables for each included study

SLR vs. Narrative Review

DimensionNarrative ReviewSLR (Kitchenham)
SearchOpportunisticSystematic, documented
InclusionAuthor’s discretionExplicit criteria, applied independently
BiasHigh (publication, selection)Controlled and documented
ReproducibilityLowHigh — protocol is pre-registered
UpdateAd hocStructured (same protocol)

PUMA Integration

  • Ch.2 Literature Review methodology section: Kitchenham & Charters (2007) as the SLR protocol basis
  • PRISMA-Log: Documents the search execution against this protocol
  • Bibliography supplement: All included papers validated per Kitchenham criteria

MOCs