Mark Burgin’s Publication Record and Collaboration Network, 1973–2025

1. Introduction

A publication record is most useful when every aggregate value can be traced to the records from which it was calculated. This requirement is especially important for a long publication history spanning books, journal articles, conference papers, chapters, abstracts, reports, encyclopedia entries, and preprints. Counts alone can conceal substantial differences in the length and temporal pattern of joint work.

The purpose of this reconstruction is documentary: to establish a bibliography for Mark Burgin and to derive co-publication statistics from it. Page counts are retained as a separate quantity rather than combined with publication counts in a synthetic score. Joint-publication span and the number of distinct publication years are also kept separate. No measure is weighted by publication type, venue, authorship position, citations, or an external platform ranking.

The resulting archive consists of a reader-oriented manuscript and workbook together with publication-level, source-level, author-link, ranking, type-distribution, and quality-control tables. Appendix C is the full reference list for the 805 publications on which the analysis is based; the same records and every ranking remain available in machine-readable supplementary data.

2. Materials and Methods

2.1. Sources and Dataset Freeze

Three complementary sources were used. The itemized IS4SI profile bibliography supplied broad historical coverage, including older, regional, and non-English items [1]. DBLP supplied structured computer-science records and stable keys [2,3]. OpenAlex supplied structured work and author identifiers and supported cross-source discovery [4,5]. All retrieved source records and snapshots were frozen together on 8 August 2026.

Table 1. Bibliographic source layer.

Source	Records	Principal role	Retained provenance
IS4SI itemized bibliography	774	Historical and non-indexed coverage	Section and item number; raw citation
DBLP	154	Structured computer-science metadata	DBLP key; title; year; pages; authors
OpenAlex	353	Discovery and cross-source matching	Work ID; DOI; authors; venue; pagination

Source counts are raw retained records, not unique publications.

Table 1. Bibliographic source layer.

Source	Records	Principal role	Retained provenance
IS4SI itemized bibliography	774	Historical and non-indexed coverage	Section and item number; raw citation
DBLP	154	Structured computer-science metadata	DBLP key; title; year; pages; authors
OpenAlex	353	Discovery and cross-source matching	Work ID; DOI; authors; venue; pagination

Source counts are raw retained records, not unique publications.

The focus was the work of UCLA-affiliated Professor Mark Burgin represented by DBLP pid b/MarkBurgin and OpenAlex author A5109006861. The 47 clusters initially found through OpenAlex received a record-level audit. Twenty-two independently attributable publications were admitted to the core. Twenty-five records were excluded with an explicit reason, such as duplicate manifestations, component chapters of an authored monograph, a pre-publication version, aggregate or review metadata that did not establish authorship, and same-name authors working in unrelated fields. All candidate and source rows remain retained.

2.2. Record Normalization and Author Identity

Each source record retained its raw citation, source identifier, source URL, retrieval date, parsed title, year, venue, type, author string, DOI or external identifier, page expression, parsed page count, and parsing flags. Titles were normalized for matching by case-folding, punctuation removal, whitespace normalization, and a small set of documented typographic corrections. The original form remained unchanged in the source layer.

Author names were normalized without deleting the source form. Full names from linked DBLP or OpenAlex records were preferred; otherwise the most specific documented initials were retained. Diacritics were preserved. Obvious spelling variants were consolidated through a documented alias table and a final near-match audit. Similar surnames with conflicting initials were kept separate. The final audit consolidated the source variants Mikkililineni/Mikkilineni, Sadovski/Sadovskii with identical initials, and Paul Zellweger/H. Paul Zellweger.

2.3. Deduplication and Version Control

Records were linked in descending order of specificity: normalized DOI equality; stable DBLP or OpenAlex identifiers; exact normalized title with compatible year; and conservative cross-source title similarity with compatible year, type, and authorship. Fuzzy links remained reviewable through method and quality fields. Source rows describing the same artifact were collapsed into one canonical record. A preprint clearly corresponding to a later formal publication was retained as a source manifestation but excluded from the core metrics. Six IS4SI conference-program abstracts were likewise retained in the audit layer but excluded because their formal Proceedings versions were already present, sometimes under expanded titles. Separately published abstracts, translations, editions, encyclopedia entries, and other distinct artifacts remained separate.

2.4. Publication Types and Pages

Publication types were descriptive and unweighted: journal article, conference paper, abstract, extended abstract when explicitly identified, book chapter, book, edited book, editorial item, encyclopedia entry, technical report, preprint, and other. An inclusive page count was calculated as end minus start plus one. Discontinuous ranges were summed, explicit stated lengths were accepted, and explicit book lengths were retained. Article numbers were not treated as pages unless a separate source stated the length. Missing pagination was stored as null, never as zero.

For a co-author, known joint pages equal the sum of the page counts of joint publications with usable pagination. If a multi-authored publication has p pages, p contributes to the known-page total of each co-author on that publication. Every page total is accompanied by the number and percentage of joint publications for which pages were known.

2.5. Publication Dates and Posthumous Records

Mark Burgin died on 18 February 2023 [6]. A publication released after that date remains part of the bibliographic record when the publisher identifies Burgin as an author, but it is marked as a posthumous publication. Three included records meet this definition: A Systematic Approach to Autonomous Agents (2024) [7], Symmetric Instruction Machines and Symmetric Turing Machines (2025) [8], and the book-chapter version of Relational Structure of Conceptual Spaces (2025) [9]. These are publication dates, not evidence of active collaboration after death.

When an exact publisher date was available it was retained with its basis. Otherwise the recorded bibliographic year remains the temporal field. The co-author statistics therefore use the terms first and last joint-publication year and joint-publication span. They do not claim to measure the start, end, continuity, or duration of interpersonal research activity.

2.6. Joint-Publication Metrics and Rankings

Table 2. Co-author metrics and separate ranking directions.

Metric	Definition	Ranking
Unique joint publications	Number of included canonical publications shared with Mark Burgin	Descending
Known joint pages	Sum of known inclusive page counts across joint publications	Descending
First joint-publication year	Earliest recorded year of a joint publication	Ascending
Last joint-publication year	Latest recorded year of a joint publication	Descending
Joint-publication span	Last publication year - first publication year + 1	Descending
Distinct publication years	Number of years containing at least one joint publication	Descending
Publications per span-year	Joint publications divided by inclusive joint-publication span	Descending
Median known pages	Median page count among joint publications with known pages	Descending
Distinct recorded venues	Number of distinct non-empty preferred venue strings	Descending

Ties use competition ranking: tied values receive the same rank and the next rank is skipped.

Table 2. Co-author metrics and separate ranking directions.

Metric	Definition	Ranking
Unique joint publications	Number of included canonical publications shared with Mark Burgin	Descending
Known joint pages	Sum of known inclusive page counts across joint publications	Descending
First joint-publication year	Earliest recorded year of a joint publication	Ascending
Last joint-publication year	Latest recorded year of a joint publication	Descending
Joint-publication span	Last publication year - first publication year + 1	Descending
Distinct publication years	Number of years containing at least one joint publication	Descending
Publications per span-year	Joint publications divided by inclusive joint-publication span	Descending
Median known pages	Median page count among joint publications with known pages	Descending
Distinct recorded venues	Number of distinct non-empty preferred venue strings	Descending

Ties use competition ranking: tied values receive the same rank and the next rank is skipped.

The collaboration network is centered on Mark Burgin. Each normalized co-author forms one node linked to the focal author; the publication-author bridge preserves the publication memberships needed to reconstruct multi-author relations. Publication-type counts were calculated separately for every co-author. No composite index was calculated.

2.7. Reproducibility

The frozen package contains the canonical publication table, all source records, the publication-author bridge, co-author summaries, nine complete ranking tables, co-author publication-type counts, the overall type distribution, quality-control flags, source snapshots, a workbook, and SHA-256 checksums. Every value in the manuscript tables is generated from these files.

2.8. AI-Assisted Tools Disclosure

OpenAI Codex, an AI-assisted coding and writing system, was used to support structured bibliographic-metadata retrieval, record parsing and normalization, the identification of candidate duplicates and author-name variants, deterministic calculations, consistency checks, manuscript drafting, and document formatting. The system itself was not treated as a bibliographic source. Source records, matching rules, adjudication fields, scripts, and publication-level outputs are retained so that its work can be checked independently. The author defined the scope, inclusion principles, metrics, and interpretive limits and remains responsible for final verification of the dataset, analysis, and manuscript.

3. Results

3.1. Reconstruction Outcome

Table 3. Reconstruction summary.

Measure	Value
Raw source records retained	1,281
Canonical publication records	837
Included in core metrics	805
Excluded canonical records	32 (25 identity/bibliographic adjudications; 6 conference-abstract versions; 1 pre-publication version)
Included posthumous publications	3
Core records with single-source documentation	538
Core records with a DOI	231
Core records with a recorded year	805 (1973-2025)
Core records with usable pagination	668 (83.0%)
Known page sum across core records	16,890
Sole-authored records	358
Co-authored records	447
Distinct normalized co-authors	160
Co-author-publication incidences	551

The known-page sum is coverage-limited and is not an estimate for missing records.

Table 3. Reconstruction summary.

Measure	Value
Raw source records retained	1,281
Canonical publication records	837
Included in core metrics	805
Excluded canonical records	32 (25 identity/bibliographic adjudications; 6 conference-abstract versions; 1 pre-publication version)
Included posthumous publications	3
Core records with single-source documentation	538
Core records with a DOI	231
Core records with a recorded year	805 (1973-2025)
Core records with usable pagination	668 (83.0%)
Known page sum across core records	16,890
Sole-authored records	358
Co-authored records	447
Distinct normalized co-authors	160
Co-author-publication incidences	551

The known-page sum is coverage-limited and is not an estimate for missing records.

The core dataset contains 805 records. Of these, 447 (55.5%) include at least one co-author and 358 (44.5%) list Mark Burgin alone. All included records have a recorded publication year.

Figure 1. Annual publication records by co-authorship status. n=805; all included records have a recorded year. Counts are unweighted bibliographic artifacts; posthumous releases are counted in their publication year.

Figure 1. Annual publication records by co-authorship status. n=805; all included records have a recorded year. Counts are unweighted bibliographic artifacts; posthumous releases are counted in their publication year.

3.2. Publication Types and Pagination

Table 4. Publication-type distribution and page-data coverage.

Publication type	Count	% core	Page-known	Coverage	Known pages	Co-authored
journal article	264	32.8%	241	91.3%	3,381	133
conference paper	205	25.5%	185	90.2%	1,142	143
abstract	131	16.3%	82	62.6%	155	58
book chapter	107	13.3%	99	92.5%	1,693	71
preprint	32	4.0%	19	59.4%	444	10
book	28	3.5%	14	50.0%	6,362	11
encyclopedia entry	15	1.9%	15	100.0%	29	2
edited book	9	1.1%	7	77.8%	3,434	8
editorial item	8	1.0%	3	37.5%	14	5
other	3	0.4%	0	0.0%	0	3
technical report	3	0.4%	3	100.0%	236	3

Book and edited-book page totals are explicit full-volume lengths when documented. Article numbers are excluded from page counts.

Table 4. Publication-type distribution and page-data coverage.

Publication type	Count	% core	Page-known	Coverage	Known pages	Co-authored
journal article	264	32.8%	241	91.3%	3,381	133
conference paper	205	25.5%	185	90.2%	1,142	143
abstract	131	16.3%	82	62.6%	155	58
book chapter	107	13.3%	99	92.5%	1,693	71
preprint	32	4.0%	19	59.4%	444	10
book	28	3.5%	14	50.0%	6,362	11
encyclopedia entry	15	1.9%	15	100.0%	29	2
edited book	9	1.1%	7	77.8%	3,434	8
editorial item	8	1.0%	3	37.5%	14	5
other	3	0.4%	0	0.0%	0	3
technical report	3	0.4%	3	100.0%	236	3

Book and edited-book page totals are explicit full-volume lengths when documented. Article numbers are excluded from page counts.

Journal articles (264; 32.8%) and conference papers (205; 25.5%) together account for 469 records. The core also contains 131 abstracts and 107 book chapters. Pagination coverage is 83.0% overall but differs by type; the missing values remain explicit in the supplement.

3.3. Collaboration Network

The focal-author network contains 160 normalized co-authors and 551 co-author-publication incidences across 447 co-authored records. 90 co-authors (56.2%) occur in one joint publication, while 24 occur in five or more. The following tables show the twenty largest unique-publication counts without combining them with any other metric.

Table 5. Co-authors ranked by unique joint publications (top 20).

Rank	Co-author	Joint pubs.	Known pages	Page-known	Coverage
1	Vladimir Kuznetsov	80	1,607	75	93.8%
2	Narayan C. Debnath	32	180	30	93.8%
3	Gordana Dodig-Crnković	31	1,979	19	61.3%
4	Rao Mikkilineni	23	145	13	56.5%
5	Eugene Eberbach	19	265	18	94.7%
6	Chelovsky, Yu.A.	14	48	11	78.6%
7	Damon Shing-Min Liu	13	282	12	92.3%
8	Walter J. Karplus	13	282	12	92.3%
9	E. S. Burgina	10	25	3	30.0%
10	Yu. M. Borodyanskii	10	58	9	90.0%
11	Kavunenko, L.F.	9	24	8	88.9%
12	Wolfgang Hofkirchner	9	1,136	7	77.8%
13	José María Díaz Nafría	8	301	7	87.5%
14	Stepenko, G.V.	8	45	8	100.0%
15	Marc L. Smith	7	151	6	85.7%
16	Parfenzeva, N.A.	7	15	6	85.7%
17	Allen Klinger	6	80	6	100.0%
18	Arkadiy Dantsker	6	36	4	66.7%
19	Oktay Duman	6	67	5	83.3%
20	Cristian S. Calude	5	556	5	100.0%

Known pages are reported separately from publication count. Coverage is the share of that co-author’s joint publications with usable pagination.

Table 5. Co-authors ranked by unique joint publications (top 20).

Rank	Co-author	Joint pubs.	Known pages	Page-known	Coverage
1	Vladimir Kuznetsov	80	1,607	75	93.8%
2	Narayan C. Debnath	32	180	30	93.8%
3	Gordana Dodig-Crnković	31	1,979	19	61.3%
4	Rao Mikkilineni	23	145	13	56.5%
5	Eugene Eberbach	19	265	18	94.7%
6	Chelovsky, Yu.A.	14	48	11	78.6%
7	Damon Shing-Min Liu	13	282	12	92.3%
8	Walter J. Karplus	13	282	12	92.3%
9	E. S. Burgina	10	25	3	30.0%
10	Yu. M. Borodyanskii	10	58	9	90.0%
11	Kavunenko, L.F.	9	24	8	88.9%
12	Wolfgang Hofkirchner	9	1,136	7	77.8%
13	José María Díaz Nafría	8	301	7	87.5%
14	Stepenko, G.V.	8	45	8	100.0%
15	Marc L. Smith	7	151	6	85.7%
16	Parfenzeva, N.A.	7	15	6	85.7%
17	Allen Klinger	6	80	6	100.0%
18	Arkadiy Dantsker	6	36	4	66.7%
19	Oktay Duman	6	67	5	83.3%
20	Cristian S. Calude	5	556	5	100.0%

Known pages are reported separately from publication count. Coverage is the share of that co-author’s joint publications with usable pagination.

Table 6. Joint-publication-year profile of the same top-20 publication-count group.

Co-author	First pub.	Last pub.	Pub. span	Pub. years	Pubs./span yr.	Venues
Vladimir Kuznetsov	1984	2001	18	13	4.44	75
Narayan C. Debnath	2003	2013	11	10	2.91	14
Gordana Dodig-Crnković	2009	2024	16	10	1.94	19
Rao Mikkilineni	2014	2022	9	9	2.56	17
Eugene Eberbach	2006	2020	15	10	1.27	16
Chelovsky, Yu.A.	1982	1990	9	5	1.56	13
Damon Shing-Min Liu	2000	2001	2	2	6.50	12
Walter J. Karplus	2000	2001	2	2	6.50	12
E. S. Burgina	1980	1983	4	4	2.50	4
Yu. M. Borodyanskii	1991	1995	5	5	2	9
Kavunenko, L.F.	1993	1996	4	3	2.25	9
Wolfgang Hofkirchner	2015	2019	5	3	1.80	4
José María Díaz Nafría	2011	2025	15	5	0.53	5
Stepenko, G.V.	1986	1998	13	5	0.62	8
Marc L. Smith	2006	2010	5	4	1.40	5
Parfenzeva, N.A.	1996	1997	2	2	3.50	4
Allen Klinger	2003	2018	16	5	0.38	5
Arkadiy Dantsker	1993	2022	30	5	0.20	5
Oktay Duman	2008	2011	4	4	1.50	5
Cristian S. Calude	2007	2017	11	4	0.45	4

Publication span is inclusive. Publication years counts distinct years containing at least one joint publication. A posthumous release year does not imply activity after death. Venues are distinct preferred venue strings.

Table 6. Joint-publication-year profile of the same top-20 publication-count group.

Co-author	First pub.	Last pub.	Pub. span	Pub. years	Pubs./span yr.	Venues
Vladimir Kuznetsov	1984	2001	18	13	4.44	75
Narayan C. Debnath	2003	2013	11	10	2.91	14
Gordana Dodig-Crnković	2009	2024	16	10	1.94	19
Rao Mikkilineni	2014	2022	9	9	2.56	17
Eugene Eberbach	2006	2020	15	10	1.27	16
Chelovsky, Yu.A.	1982	1990	9	5	1.56	13
Damon Shing-Min Liu	2000	2001	2	2	6.50	12
Walter J. Karplus	2000	2001	2	2	6.50	12
E. S. Burgina	1980	1983	4	4	2.50	4
Yu. M. Borodyanskii	1991	1995	5	5	2	9
Kavunenko, L.F.	1993	1996	4	3	2.25	9
Wolfgang Hofkirchner	2015	2019	5	3	1.80	4
José María Díaz Nafría	2011	2025	15	5	0.53	5
Stepenko, G.V.	1986	1998	13	5	0.62	8
Marc L. Smith	2006	2010	5	4	1.40	5
Parfenzeva, N.A.	1996	1997	2	2	3.50	4
Allen Klinger	2003	2018	16	5	0.38	5
Arkadiy Dantsker	1993	2022	30	5	0.20	5
Oktay Duman	2008	2011	4	4	1.50	5
Cristian S. Calude	2007	2017	11	4	0.45	4

Publication span is inclusive. Publication years counts distinct years containing at least one joint publication. A posthumous release year does not imply activity after death. Venues are distinct preferred venue strings.

Table 7. Publication-type distribution for the ten largest joint-publication counts.

Co-author	Journal	Conf.	Abstract	Chapter	Book/ed.	Preprint	Other
Vladimir Kuznetsov	18	23	6	29	4	0	0
Narayan C. Debnath	3	28	1	0	0	0	0
Gordana Dodig-Crnković	10	4	2	5	4	2	4
Rao Mikkilineni	10	7	2	0	0	3	1
Eugene Eberbach	11	5	1	2	0	0	0
Chelovsky, Yu.A.	3	6	1	4	0	0	0
Damon Shing-Min Liu	2	8	1	0	0	0	2
Walter J. Karplus	2	8	1	0	0	0	2
E. S. Burgina	2	0	8	0	0	0	0
Yu. M. Borodyanskii	5	1	3	1	0	0	0

Other combines editorial items, encyclopedia entries, technical reports, and records classified as other. Complete per-co-author distributions are in Supplement S6.

Table 7. Publication-type distribution for the ten largest joint-publication counts.

Co-author	Journal	Conf.	Abstract	Chapter	Book/ed.	Preprint	Other
Vladimir Kuznetsov	18	23	6	29	4	0	0
Narayan C. Debnath	3	28	1	0	0	0	0
Gordana Dodig-Crnković	10	4	2	5	4	2	4
Rao Mikkilineni	10	7	2	0	0	3	1
Eugene Eberbach	11	5	1	2	0	0	0
Chelovsky, Yu.A.	3	6	1	4	0	0	0
Damon Shing-Min Liu	2	8	1	0	0	0	2
Walter J. Karplus	2	8	1	0	0	0	2
E. S. Burgina	2	0	8	0	0	0	0
Yu. M. Borodyanskii	5	1	3	1	0	0	0

Other combines editorial items, encyclopedia entries, technical reports, and records classified as other. Complete per-co-author distributions are in Supplement S6.

3.4. Separate Metric Rankings

A co-author’s position differs across publication count, known page volume, joint-publication span, publication-year coverage, rate, median length, and venue count. These dimensions are therefore not merged. Appendix A reports ranks 1-5 for every metric, retaining all ties. Supplement S5 contains the complete ranking of all co-authors for each metric, and Supplement S4 contains the underlying values, page-coverage fields, and posthumous joint-publication counts.

4. Scope and Limitations

• The IS4SI bibliography is extensive but contains inconsistent punctuation, incomplete years or pages, and occasional category or spelling variation. Single-source records are documented as such rather than promoted to multi-source verification.
• DBLP and OpenAlex do not cover all older, non-English, regional, book, abstract, or encyclopedia material. Index absence is therefore not treated as evidence of non-publication.
• 0 included records lack a year and 137 lack usable pagination. Page totals and page-based rankings are explicitly coverage-limited.
• 3 records were published after Burgin’s death on 18 February 2023. Their publication years extend the bibliographic record but are not interpreted as years of active collaboration.
• Books and edited books contribute their documented full length to a co-author’s known joint pages. This is a bibliographic page measure and does not allocate pages among editors or authors.
• Distinct venues are counts of preferred recorded venue strings. They are not a fully authority-controlled journal or conference series registry.
• The network is centered on one focal author. It describes documented co-publication relations and does not measure contribution, intellectual proximity, influence, or interpersonal ties.
• The dataset is frozen at 8 August 2026.

5. Data Availability

The section headed Data sources and methodological references identifies the bibliographic services, methods literature, memorial record, and publisher records cited in the methods. Appendix C is the complete analytic reference corpus: it reproduces every one of the 805 publications counted or classified in the analysis and links each entry to its stable publication ID. Appendix D lists every excluded canonical candidate and its exclusion reason. The preprint package also includes the complete canonical table (Supplement S1), raw source records (S2), publication-author bridge (S3), co-author metric table (S4), nine full ranking tables (S5), per-co-author publication-type distributions (S6), the overall type distribution (S7), quality-control records (S8), the reader-oriented workbook (S9), the reconstruction protocol (S10), the dataset manifest (S11), and the source snapshots. File-level SHA-256 checksums are included in the package manifest. These files are the authoritative basis for reproducing every table and appendix entry.

6. Conclusion

The reconstruction establishes a frozen, publication-level archive of 805 included bibliographic records and a co-publication table for 160 normalized co-authors. Counts, known pages, publication types, joint-publication dates, inclusive publication spans, distinct publication years, rates, median lengths, posthumous-release counts, and recorded venue counts remain separate descriptive measures. The full analytic reference list in Appendix C and the retained source layer support checking and later correction without requiring an interpretive or composite assessment.