ACM CCS Concepts: Classification Guide for Authors and acmart Submissions

CCS concepts (Computing Classification System concepts) are standardized subject taxonomy terms maintained by the Association for Computing Machinery (ACM) to categorize and index scholarly literature in the computing disciplines. When authors prepare manuscripts for ACM conference proceedings or journals using the official acmart LaTeX document class, they must assign relevant CCS concepts to index their paper’s core themes. Within academic computer science, this indexing structure determines how digital repositories, search engines, and bibliographic databases organize and retrieve research papers.

Selecting appropriate CCS concepts ensures that a manuscript reaches peer reviewers, citation indices, and domain researchers searching the ACM Digital Library. Understanding how to navigate the 2012 classification ontology, assign significance weights, and format the necessary XML and LaTeX commands simplifies submission for both conference and journal tracks.

What Are CCS Concepts?

The ACM Computing Classification System originated in 1964 as a mechanism to organize technical literature published across ACM periodicals and conference proceedings. For decades, the system relied on an alphanumeric tree structure, culminating in the widely used 1998 revision where topics received four-level alphanumeric codes such as H.3.3 (Information Search and Retrieval) or I.2.6 (Learning).

In 2012, ACM completely overhauled the system, replacing the rigid alphanumeric notation with a poly-hierarchical semantic ontology. Under the 2012 CCS ontology, individual concepts can possess multiple parent categories, reflecting the interdisciplinary nature of contemporary computing. Each concept has a permanent numerical concept identifier (for instance, 10010147.10101784) alongside a human-readable descriptive hierarchy path separated by tildes (~).

Who Maintains the CCS and What It Covers

The classification scheme is developed, governed, and updated by the Association for Computing Machinery. It provides a shared taxonomy across computing subfields, guiding conference program committees, editorial boards, and digital catalogers.

The 2012 ACM Computing Classification System spans 13 first-level root categories:

• General and reference: Document types, surveys, cross-computing tools, and general empirical principles.
• Hardware: Circuit design, integrated circuits, power analysis, communications hardware, and emerging architectures.
• Computer systems organization: Embedded and cyber-physical systems, real-time architectures, distributed computing, and cloud systems.
• Networks: Network protocols, network architecture, security architectures, and physical link implementations.
• Software and its engineering: Software development methodologies, formal methods, operating systems, compilers, and programming languages.
• Theory of computation: Computability, computational complexity, formal languages, algorithmic analysis, and cryptography.
• Mathematics of computing: Discrete mathematics, probability, numerical analysis, mathematical optimization, and graph algorithms.
• Information systems: Database management systems, search engines, information retrieval, and data mining architectures.
• Security and privacy: Cryptographic protocols, intrusion detection, software security, network security, and human factors in security.
• Human-centered computing: Human-computer interaction (HCI), collaborative systems, accessibility, and visual analytics.
• Computing methodologies: Artificial intelligence, machine learning, computer graphics, computer vision, and simulation.
• Applied computing: Enterprise computing, electronic commerce, physical sciences, life sciences, and law.
• Professional topics: Computing profession ethics, computing education, intellectual property, and computing history.

How to Select and Assign CCS Concepts

ACM authors select concepts through the interactive CCS Concept Generator hosted on the ACM Digital Library platform (dl.acm.org/ccs). When assigning terms, authors follow five primary steps:

1. Access the taxonomy tool: Navigate to the web-based ACM CCS Concept Generator to explore the classification tree.
2. Search for primary terms: Enter research keywords into the search filter or manually expand the relevant first-level domains to locate leaf nodes that match the paper’s methodology and technical contributions.
3. Assign relevance weights: Assign each selected concept an explicit significance rating based on how central the subject is to the paper:
   • 500 (High / Primary relevance): Represents the core technical focus or novel contribution of the article. Most papers select one or two primary concepts.
   • 300 (Medium relevance): Represents important secondary techniques, secondary evaluations, or architectural dependencies.
   • 100 (Low relevance): Represents background context, peripheral application domains, or supplementary methodologies.
4. Review assigned concepts: Ensure that the selected concepts balance technical depth with discoverability, typically totaling between two and six concepts per article.
5. Generate code snippets: Click the generation button to create both the machine-readable XML block and the corresponding LaTeX typesetting commands.

Implementing CCS Concepts in acmart

Authors writing in LaTeX using the official acmart document class must embed both the raw XML classification block and the typesetting macro before issuing the maketitle command. The XML snippet is enclosed in a CCSXML environment, followed by individual ccsdesc macro calls specifying numerical significance weights:

begin{CCSXML}
<ccs2012>
  <concept>
    <concept_id>10010147.10101784.10101785</concept_id>
    <concept_desc>Computing methodologies~Machine learning</concept_desc>
    <concept_significance>500</concept_significance>
  </concept>
  <concept>
    <concept_id>10002978.10003022.10003028</concept_id>
    <concept_desc>Security and privacy~Domain-specific security and privacy architectures</concept_desc>
    <concept_significance>300</concept_significance>
  </concept>
</ccs2012>
end{CCSXML}

ccsdesc[500]{Computing methodologies~Machine learning}
ccsdesc[300]{Security and privacy~Domain-specific security and privacy architectures}

According to the official acmart documentation, CCS concepts and user-defined keywords are mandatory for all regular research papers exceeding two pages in length. For one-page or two-page extended abstracts, posters, and short summaries, CCS concepts are optional. By default, the template automatically prints the concepts on the opening page; authors can regulate this behavior through the settopmatter{printccs=true} parameter.

Even if an article includes translated abstracts or keywords in languages other than English, CCS concepts are always typeset and archived exclusively in English. Authors should not confuse CCS concepts with freeform author keywords defined via the keywords{...} macro. Freeform keywords supply granular, unstandardized terms, whereas CCS concepts establish fixed ontological links within digital libraries.

When coordinating publication workflows across collaborative projects, authors should also align their submissions with the ACM Policy on Authorship and review opportunities for software or dataset verification via ACM Artifact Review and Badging, particularly when submitting to prestigious tracks in the field of computer science conference and journal publishing.

Strengths and Limitations of the System

The 2012 revision brought structural improvements over earlier indexing iterations, but also carries distinct operational boundaries:

• Strengths:
  • Poly-hierarchical relationships: Research intersecting multiple domains (such as cryptographic hardware or machine learning for wireless networks) can be indexed under multiple parent categories without arbitrary compartmentalization.
  • Automated ingestion: The dual XML and LaTeX implementation allows automated harvesting tools to populate the ACM Digital Library catalog without manual data entry.
  • Uniform discoverability: Replaces idiosyncratic author tagging with a controlled vocabulary vetted by computing professionals.
• Limitations:
  • Static ontology snapshot: Because the ontology dates to 2012, highly contemporary subfields (such as large language models, prompt tuning, or quantum machine learning) lack dedicated leaf nodes, requiring authors to map them to broader parent nodes like Computing methodologies~Artificial intelligence.
  • Formatting overhead: Mismatches between the XML tree IDs and the human-readable ccsdesc strings in the LaTeX source can cause compilation warnings or metadata ingestion rejections during publisher production checks.

CCS Concepts vs. Alternative Classification Schemes

Different scientific societies and scholarly publishing organizations maintain dedicated taxonomies to organize technical literature:

Classification Scheme	Governing Body	Ontological Structure	Primary Discipline Focus	Integration Format
ACM CCS (2012)	Association for Computing Machinery (ACM)	Poly-hierarchical semantic ontology (13 roots)	Computer science and applied computing	Embedded XML metadata and ccsdesc macros in acmart
PhySH (Physics Subject Headings)	American Physical Society (APS)	Faceted poly-hierarchical classification	Physics and astronomy	Web-based submission portal selection
IEEE Taxonomy	Institute of Electrical and Electronics Engineers (IEEE)	Hierarchical controlled thesaurus	Electrical engineering, electronics, and computing	Keywords listed in manuscript header
Mathematics Subject Classification (MSC2020)	AMS and zbMATH	Alphanumeric 5-character hierarchy (63 top-level areas)	Pure and applied mathematics	LaTeX subjclass macro in AMS-LaTeX packages

Frequently Asked Questions

Are CCS concepts mandatory for all ACM submissions?

Yes, for full research articles. The official acmart class specification mandates CCS concepts and author keywords for all submissions exceeding two pages. They are optional for one- and two-page articles, posters, and short abstracts.

What do the significance numbers 500, 300, and 100 mean?

The numbers indicate topical relevance in descending order: 500 signifies primary topic relevance (the paper’s central contribution), 300 denotes medium relevance (secondary methods or major components), and 100 indicates low relevance (contextual applications or peripheral concepts).

Where should the CCSXML block be placed in a LaTeX file?

The begin{CCSXML} ... end{CCSXML} environment and accompanying ccsdesc macro commands must appear in the document preamble or top-matter group, placed strictly before the maketitle command.

How do CCS concepts differ from author keywords?

CCS concepts are standardized terms drawn from a curated, hierarchical ontology that ACM uses for structured metadata, faceted filtering, and citation indexing. Author keywords (entered using keywords{...}) are unstandardized, free-text terms chosen by authors to capture specific keywords, acronyms, or emerging terms not yet present in the taxonomy.

Sources

• Boris Veytsman. The acmart Class: User’s Guide for the ACM Master Article Template. Comprehensive TeX Archive Network (CTAN), mirrors.ctan.org/macros/latex/contrib/acmart/acmart.pdf (read 2026-11-07).
• TaxoBank. ACM Computing Classification System (2012 Edition). taxobank.org/vocabularies/acm-computing-classification-system-2012 (read 2026-11-07).