Chapter 6: Literature Review¶
🎓 Learning Objectives
- Understand the purpose of literature review
- Learn systematic literature review process
- Master search strategies
- Learn to synthesize and analyze papers
- Understand how to write literature review sections
What is a Literature Review?¶
A literature review is a comprehensive survey of existing research in your area. It:
- Summarizes existing work
- Identifies gaps and opportunities
- Positions your research
- Provides theoretical foundation
- Shows your understanding of the field
Literature Review Importance
A good literature review: - Demonstrates your knowledge - Justifies your research - Helps avoid duplication - Identifies research gaps - Guides your research direction
Types of Literature Reviews¶
1. Narrative Review¶
Purpose: Broad overview of field
Characteristics: - Thematic organization - Qualitative synthesis - Broad coverage - Less systematic
Use: General understanding, introduction sections
2. Systematic Review¶
Purpose: Comprehensive, systematic analysis
Characteristics: - Systematic search - Explicit criteria - Quantitative synthesis - Reproducible
Use: Comprehensive analysis, survey papers
Systematic Reviews
More rigorous but time-consuming. Use for important reviews.
3. Meta-Analysis¶
Purpose: Statistical synthesis of results
Characteristics: - Quantitative analysis - Statistical methods - Effect size calculation - Comprehensive
Use: When multiple studies exist on same topic
Literature Review Process¶
Step 1: Define Scope¶
Questions to Answer: - What is the research question? - What time period to cover? - What types of papers? - What languages? - What venues?
Scope Definition
Start broad, then narrow. Better to include more initially.
Step 2: Search Strategy¶
Search Components: - Keywords: Main terms - Databases: Where to search - Filters: Time, type, language - Boolean operators: AND, OR, NOT
Example Search:
("object detection" OR "object recognition")
AND ("deep learning" OR "neural network")
AND ("small objects" OR "tiny objects")
Search Iteration
Refine searches based on results. Try different keyword combinations.
Step 3: Paper Collection¶
Sources: - Academic databases - Conference proceedings - Journal articles - Preprint servers - Citation chains
Tools: - Reference managers (Zotero, Mendeley) - Search engines (Google Scholar, Semantic Scholar) - Specialized platforms (Papers With Code)
Paper Collection
- Don't collect too many initially
- Focus on quality over quantity
- Use reference managers
- Organize systematically
Step 4: Screening¶
Criteria: - Relevance to research question - Quality of work - Recency (if relevant) - Venue quality
Process: 1. Title/abstract screening 2. Full-text screening 3. Final selection
Screening Process
- Use inclusion/exclusion criteria
- Document decisions
- Get second opinion if unsure
- Keep rejected papers for reference
Step 5: Reading and Analysis¶
For Each Paper: - Read carefully - Take detailed notes - Extract key information - Identify contributions - Note limitations
Information to Extract: - Problem addressed - Method proposed - Key contributions - Results achieved - Limitations - Future work
Step 6: Synthesis¶
Organize by: - Chronological order - Thematic groups - Methodological approaches - Problem domains
Synthesis Tasks: - Identify themes - Compare approaches - Find patterns - Identify gaps - Note contradictions
Synthesis Tips
- Group related papers
- Identify trends
- Note evolution of ideas
- Find common themes
Step 7: Writing¶
Structure: 1. Introduction to area 2. Thematic sections 3. Comparison and analysis 4. Gaps and opportunities 5. Summary
Search Strategies¶
Keyword Development¶
Process: 1. Start with main terms 2. Find synonyms 3. Use controlled vocabularies 4. Check paper keywords 5. Refine iteratively
Sources for Keywords: - Your research question - Related papers - Subject headings - Thesauri - Expert input
Keyword Tips
- Use both specific and general terms
- Include acronyms
- Try different phrasings
- Use Boolean operators
Database Selection¶
Primary Databases: - Google Scholar - Semantic Scholar - arXiv - Conference proceedings - Journal databases
Specialized: - Papers With Code (for implementations) - DBLP (computer science) - ACM Digital Library - IEEE Xplore
Multiple Databases
Search multiple databases. Each has different coverage.
Citation Chaining¶
Forward Chaining: - Papers that cite your paper - Shows influence and extensions
Backward Chaining: - Papers cited by your paper - Shows foundation and related work
Tools: - Google Scholar "Cited by" - Semantic Scholar citations - Connected Papers
Citation Chaining
Excellent way to find related work. Follow citation chains systematically.
Organizing Literature¶
Thematic Organization¶
Structure:
Literature Review
├── Introduction
├── Theme 1: Method A
│ ├── Paper 1
│ ├── Paper 2
│ └── Comparison
├── Theme 2: Method B
│ ├── Paper 3
│ ├── Paper 4
│ └── Comparison
├── Theme 3: Applications
└── Gaps and Opportunities
Chronological Organization¶
Structure:
Literature Review
├── Early Work (2010-2015)
├── Recent Advances (2016-2020)
├── Current State (2021-present)
└── Trends and Directions
Methodological Organization¶
Structure:
Literature Review
├── Supervised Learning Approaches
├── Unsupervised Learning Approaches
├── Reinforcement Learning Approaches
└── Hybrid Approaches
Organization Choice
Choose organization that best serves your research. Thematic is most common.
Analysis and Synthesis¶
Comparing Papers¶
Comparison Dimensions: - Problem addressed - Method used - Datasets evaluated - Results achieved - Limitations - Computational cost
Comparison Table:
| Paper | Method | Dataset | Accuracy | Limitations |
|---|---|---|---|---|
| Paper A | Method X | Dataset 1 | 85% | High compute |
| Paper B | Method Y | Dataset 1 | 82% | Limited scale |
Comparison Tables
Useful for organizing information and identifying patterns.
Identifying Gaps¶
Gap Types: - Methodological: Missing approaches - Empirical: Unexplored datasets/problems - Theoretical: Unanswered questions - Application: Unexplored domains
Questions to Ask: - What hasn't been done? - What limitations exist? - What questions remain? - What domains unexplored?
Gap Identification
Gaps justify your research. Clearly identify and articulate them.
Trend Analysis¶
Identify: - Evolution of methods - Performance improvements - Shifts in focus - Emerging areas
Visualization: - Timeline diagrams - Performance plots - Citation networks
Writing Literature Review¶
Structure¶
1. Introduction - Scope and purpose - Organization overview - Key themes
2. Body Sections - Thematic organization - Detailed analysis - Comparisons - Critical evaluation
3. Synthesis - Summary of findings - Identified gaps - Research opportunities - Your contribution positioning
Writing Tips¶
Writing Guidelines
- Be critical: Don't just summarize
- Be organized: Clear structure
- Be comprehensive: Cover important work
- Be current: Include recent work
- Be balanced: Fair evaluation
Common Mistakes¶
Avoid These
- Just listing papers: Need analysis
- No organization: Random order
- Missing recent work: Outdated
- No gaps identified: Missing purpose
- Biased selection: Cherry picking
- Poor writing: Unclear, verbose
Literature Review Tools¶
Reference Management¶
Zotero: - Free, open-source - Browser integration - PDF management - Citation generation
Mendeley: - PDF annotation - Social features - Reference sharing
Note-Taking¶
Notion: - Flexible databases - Templates - Collaboration
Obsidian: - Markdown-based - Graph view - Local files
Visualization¶
Connected Papers: - Paper graphs - Related papers - Citation networks
VOSviewer: - Citation network visualization - Co-occurrence analysis
Resources¶
📚 Literature Review Guides
- How to Write a Literature Review - UNC Writing Center
- Systematic Reviews Guide - University of Toronto
- Literature Review Tutorial - Video guide
🛠️ Tools
- Zotero - Reference management
- Connected Papers - Paper graphs
- Semantic Scholar - AI-powered search
- ResearchRabbit - Paper discovery
📖 Examples
- Review recent survey papers in your area
- Check "Related Work" sections of top papers
- Read systematic reviews
- Study well-written reviews
Next Steps¶
- Chapter 7: Research Design & Experimental Setup - Design experiments
- Chapter 8: Data Collection & Management - Handle research data
Key Takeaways: - Literature review is comprehensive survey of existing work - Systematic process: Define scope → Search → Collect → Screen → Analyze → Synthesize → Write - Use multiple search strategies and databases - Organize thematically, chronologically, or methodologically - Identify gaps and position your research - Write critically and comprehensively