bugraayantr/disinformation-research-gaps

Disinformation Research Gap Atlas

     

An evidence-grounded atlas of research gaps in the global disinformation literature — 295 concrete gaps distilled from 40,989 English-language scholarly works across 102 thematic clusters.

Source: OpenAlex metadata (title + abstract) from the global disinformation & misinformation literature..

81% of gaps (240/295) sit on rising clusters — the field is expanding fast, but critical questions remain open.

At a Glance

🧩 Research gaps	295
🗺️ Topic clusters	102
📚 Works analyzed	40,989
📈 Rising-trend gaps	240
🎯 Avg. confidence	0.81

Dataset Preview

Browse the Dataset Viewer tab (spreadsheet-like):

• gaps — 295 gaps: topic, gap_statement_en, gap_type, research_questions (list), confidence …
• clusters — 102 clusters: topic, size, gap_count, year_trend
• gaps_csv / clusters_csv — CSV exports for Excel

Sample Gaps

Disinformation, Social Media, Fake News, Digital · empirical · confidence 0.85

There is a need for more research to understand the impacts of fake news.

_{Sahte haberlerin etkilerini anlamak için daha fazla araştırma yapılması gerekmektedir.}

Research questions:
• Sahte haberlerin demokrasi üzerindeki etkileri nelerdir?
• Sahte haberlerin toplumda güveni nasıl etkilediği üzerine ne tür veriler toplanabilir?

Evidence (author statements): evaluate the impact of disinformation on democracy and contribute to the design of a policy against it.

_{1,246 supporting works · trend: rising}

Automated Fact, Fact Checking, Verification, Knowledge · methodological · confidence 0.9

There is a need for further research on the ability of automated fact-checking systems to effectively retrieve evidence from complex structured academic papers.

_{Otomatik gerçek kontrol sistemlerinin karmaşık yapılandırılmış akademik belgelerden etkili bir şekilde kanıt toplama yeteneği üzerine daha fazla araştırma gerekmektedir.}

Research questions:
• Otomatik gerçek kontrol sistemleri, karmaşık akademik belgelerden nasıl daha etkili kanıt toplayabilir?
• Mevcut sistemlerin karmaşık belgelerdeki bilgileri nasıl işlediği ve bu süreçte karşılaştıkları zorluklar nelerdir?

Evidence (author statements): This perspective paper aims to advance scientific fact-checking with a specialised IR system tailored for real-world applications.

_{742 supporting works · trend: rising}

Automated Fact, Fact Checking, Verification, Knowledge · empirical · confidence 0.85

Further research should be conducted on the potential of automated fact-checking systems to improve performance with multilingual datasets.

_{Otomatik gerçek kontrol sistemlerinin çok dilli veri setleri ile performansını artırma potansiyeli üzerine daha fazla araştırma yapılmalıdır.}

Research questions:
• Çok dilli veri setleri kullanarak otomatik gerçek kontrol sistemlerinin doğruluk oranı nasıl artırılabilir?
• Farklı dillerdeki içeriklerin otomatik gerçek kontrol sistemleri üzerindeki etkileri nelerdir?

Evidence (author statements): Using articles in a different language from the claim slightly lowers performance.

_{742 supporting works · trend: rising}

Method

Research gaps are derived from authors' own statements of limitations and future work in abstracts — an evidence-grounded pipeline:

1. Harvest — 40,989 global disinformation works (title + abstract) from OpenAlex.
2. Embedding — abstracts vectorized via LLM embeddings.
3. Clustering — 102 topic clusters (MiniBatchKMeans) + TF-IDF labeling + year trends.
4. Signal extraction — per-article LLM extraction of stated limitations, future work, methods, and concepts.
5. Gap synthesis — cluster-level LLM synthesis into concrete gaps, each with supporting work IDs and quotes.

Gap Types

Type	Count
empirical	98
methodological	96
theoretical	58
interdisciplinary	22
population	19
temporal	2

Schema — gaps

Column	Description
gap_id	unique identifier
cluster_id, topic	cluster link + cleaned topic label
gap_statement_en / gap_statement_tr	gap statement (EN/TR)
gap_type	empirical / methodological / theoretical / population / temporal / interdisciplinary
research_questions	suggested research questions (list)
evidence_article_ids	supporting {src_short} work IDs (list)
supporting_quotes	author-stated limitation / future-work quotes
n_related, year_trend	cluster size, trend (rising/saturated/declining)
confidence	model confidence score (0–1)

clusters subset: topic, label_raw, size, gap_count, year_trend.

Usage

from datasets import load_dataset
gaps = load_dataset("bugraayantr/disinformation-research-gaps", "gaps", split="train")
for g in gaps:
    if g["gap_type"] == "empirical" and g["year_trend"] == "yükselen":
        print(g["gap_statement_en"], g["research_questions"])

Limitations & Ethics

• Title + abstract only — future-work sections often live in full text; some gaps may read broadly.
• Scope is limited to the global disinformation and misinformation literature. A gap filled elsewhere may still be open here.
• Gaps are machine-generated — expert validation recommended. Every record carries evidence traces (evidence_article_ids, supporting_quotes).

Citation

@dataset{academic_gaps_2026,
  title     = {{Disinformation Research Gap Atlas}},
  author    = {{Ayan, Buğra}},
  year      = {2026},
  publisher = {{Hugging Face}},
  url       = {{https://huggingface.co/datasets/bugraayantr/disinformation-research-gaps}}
}

Source content harvested from OpenAlex küresel dezenformasyon/mis-enformasyon literatürü meta-verisi (başlık + özet).