How to write a conference paper: format and tips

Editage describes a conference paper as a short, peer-reviewed report of original research, published in the conference proceedings. It is usually presented at the conference too. In engineering and computer science it is often a student's first publication.

Conferences ask for different kinds of papers, and the call for papers will say which ones it accepts. Editage lists four common types:

Conference paper vs journal paper. A journal paper is longer and more complete, and its review is usually slower and more thorough. A conference paper is shorter, it is reviewed more quickly, and it is built to be presented in a few minutes.

Ritesh's call asked for full papers, so he needed real results, not a plan, in a limited number of pages.

The call for papers, often shortened to CFP, is the conference's rule book. Editage warns that going over the page limit "can lead to automatic rejection", so read the call before you plan a single section. Ritesh printed his and marked these points:

• The scope and tracks. Does your topic fit one of the themes the conference lists?
• The paper type and page limit. His call allowed 12 pages, including references.
• The template. His conference wanted Springer's LNCS template, in Word or LaTeX.
• The kind of review. His was double-blind, so his name could not appear in the paper.
• The deadlines. Paper submission, notification of acceptance, camera-ready version and registration.
• Where the proceedings will be published, and whether at least one author must register and present.

Then check the conference itself. Library guides on predatory conferences warn that fake ones mostly target early career researchers, and some are cancelled without refunds. The Think. Check. Attend. initiative publishes a checklist that "aims to guide and assist researchers and scholars to judge the legitimacy and academic credentials of conferences." Its questions include "Are you aware of the society or the association organizing this conference?" and "Is the Organizing Committee clear about where the proceedings will be published?"

If the conference promises Scopus indexing, check the book series behind it, not the logo on the poster. Our guide on how to check a Scopus indexed journal has a section on conferences.

Many first conference papers go wrong the way Ritesh's first draft did: they try to carry the whole thesis. Reviewers read many papers for one conference and need to see quickly what is new in yours. Guides on conference writing suggest choosing a single core contribution, such as a new method, an improvement on an existing approach, insights from a case study or a new dataset or tool, and stating it in one sentence.

Ritesh's dissertation had four parts: a review of dust monitoring methods, a survey of village health clinics, the sensor comparison and the correction model. Dr. Kerketta asked him which part another researcher could not find anywhere else. It was the correction model. After three tries his sentence read: a simple correction model makes cheap dust sensors track a monitoring station closely enough to warn nearby villages.

1. Write the sentence first.If you cannot state your contribution in one sentence, the reviewer will not be able to either.
2. Keep only what proves it.Every figure, table and paragraph should support that sentence. Ritesh kept the sensor comparison and the model, and saved the survey for later.

A conference paper follows the order of a journal paper, only shorter. Editage says most are 4 to 8 pages in two columns, and gives a typical length for each part. Here it is beside what Ritesh wrote.

Many conference papers add a short related work section after the introduction, and many combine results and discussion to save space. Follow recent papers from the same conference.

The abstract matters more than its size suggests, because committees often sort papers by it. Our guide on how to write an abstract shows how to fit problem, method, result and meaning into a few sentences. Editage also suggests reading three to five accepted papers from the conference before you write, to see the tone and the level of detail reviewers expect.

Conferences publish their papers through a publisher, and the publisher sets the layout. Editage names the common templates: IEEEtran for IEEE, the ACM template, LNCS for Springer and Elsevier's article template, each in LaTeX and Word. IEEE offers its conference template in A4 and US Letter sizes. Download the template from the conference website, since some conferences change it slightly.

Write inside the template from the start. Ritesh wrote in plain Word and moved his text into the LNCS file a week before the deadline. It ran to fifteen pages, and cutting three taught him these rules.

• Check whether references count. Limits differ. The 2024 ISSEP conference, which used LNCS, set a "strict: 12 pages, including references" limit. Some IEEE conferences count references and some do not.
• Do not shrink the margins or fonts. Some conferences that use LNCS warn that changes to the style file will be undone, and IEEE asks authors to delete the template's guidance text before submitting.
• Cut words, not results. Ritesh removed his history of dust monitoring, merged two tables into one, and shortened every figure caption.

His conference also used double-blind review, so reviewers would not see who wrote the paper. Anonymising takes more than deleting your name. Guidelines from the HRI conference ask authors to remove names and institutions from the first page, the acknowledgements and the PDF file's metadata. The SC conference asks authors to refer to their own earlier work in the third person. Ritesh changed "the dust survey at our institute" to "a mining area in eastern India", and cited Dr. Kerketta's earlier study as "Kerketta et al. [7] showed", not "our earlier work".

Conferences take papers through an online system such as EasyChair or Microsoft CMT. You fill in the title, abstract and authors, and upload the PDF. Ritesh submitted two days early and checked that the uploaded PDF showed no names.

Seven weeks later the decision came: accepted, with comments from two reviewers. Editage says acceptance rates are about 15 to 25 percent at highly competitive conferences and above 50 percent at smaller or regional ones.

An accepted paper is not finished yet. The final version you send for publication is called the camera-ready paper, and it comes with its own list:

1. Answer the reviewers.Fix what they asked for. One of Ritesh's reviewers wanted to know how the model behaved during the monsoon, so he added a short paragraph and one line to a table.
2. Put the names back.Add the authors, affiliations and acknowledgements removed for review. ISSEP 2024 also asked for each author's ORCID.
3. Sign the copyright or licence form.Springer conferences ask the corresponding author to sign a licence to publish for all the authors, and other publishers have their own forms. ISSEP noted that authorship cannot be changed once files go to Springer.
4. Register and present.Many conferences publish only papers that are presented. ISSRE 2026, an IEEE conference, says papers not presented are dropped from the proceedings and not sent to IEEE Xplore.

For the talk itself, our guide on how to present a research paper covers slides, timing and questions.

Often, yes. Many journals accept an extended version of a conference paper, but only if it adds substantial new work. The IEEE Communications Society asks for "a significant amount of new material beyond that of the conference paper(s)". It does not set a number, and it says "we do not have a quantitative threshold for determining it." Authors must "inform the editor about these prior publications" and "cite the prior conference papers".

Other editors use a rule of thumb. The editors of the journal Software Testing, Verification and Reliability describe a common guideline that a new paper should have at least 30 percent significant new material. Rules differ by publisher and journal, so read the policy of the journal you are aiming at.

New material means new work, such as more data or a new experiment, not the same results in longer sentences. Ritesh plans to add a second year of readings and a test of the model at two more mines. He will cite his conference paper and mention it in his cover letter.

Ritesh presented his paper in Nagpur. A researcher from a pollution control board asked whether the sensors could be placed near schools. He did not know yet, and said so. It became the first line of his plan for the journal version.