This blog post is part of a series of posts aimed toward supporting Diversity & Inclusion at the ISMIR2021 conference. It was authored by the Diversity & Inclusion Chairs and Scientific Program Chairs.
Springtime tends to be a busy and productive time for Music Information Retrieval (MIR) researchers, many of whom are hard at work preparing paper submissions for the annual International Society for Music Information Retrieval (ISMIR) Conference. This year's deadlines are fast approaching, with abstracts due by May 8, 2021 and full papers due by May 15, 2021. For newer researchers, or those who are new to ISMIR, this process can feel somewhat mysterious. Here we outline key aspects of the ISMIR paper writing and review process, with the aim of positioning ISMIR researchers for successful paper contributions in 2021.
Preparing and writing an ISMIR paper
The Call for Papers for ISMIR2021, along with an extensive list of Topics of Interest, can be found at https://ismir2021.ismir.net/cfp/.
ISMIR Paper Submissions Specifications
This is a summary of the submission requirements. Please view the official Call for Papers for the complete description of requirements.
ISMIR paper submissions are full papers, up to 6 pages of content, representing completed work. Works-in-progress, or papers with results that are forthcoming or not fully reported, are not suitable as full paper submissions to the ISMIR conference (but see "Other options", below).
The page limit of ISMIR paper submissions is "6+n" pages. This means that the scientific content of the paper (text, figures, and tables) must fit in no more than 6 pages. Following that, any number of additional optional pages can contain only references (acknowledgments may also be included in these later pages but should not be included in the initial paper submission, to preserve anonymity). Papers must be submitted in PDF format using the ISMIR2021 paper template, which is available in LaTeX and Word formats.
The content of ISMIR submissions must represent original contributions , which means that the research cannot have been published elsewhere previously, and cannot be currently submitted for publication at any other venue. Due to the double-blind review process (see below), it is critical that anonymity be preserved in the paper submissions. Therefore, authors should retain the anonymous author names in the paper template, should not include any links that can identify the authors, and should self-cite in the third person. Authors are also strongly discouraged from posting their submitted manuscripts (e.g., on personal websites or preprint servers), or promoting their work publicly (e.g., on social media or blogs) while their paper is under review.
Papers are submitted via the Microsoft CMT system, which is currently open for submissions. Authors must initiate their submission and provide an initial abstract by the end of May 8, 2021, Anywhere on Earth (AoE — i.e., 23:59 in the latest timezone on Earth). Following this, the authors are welcome to continue editing their submission through May 15, 2021 (AoE). In the CMT system, authors should also be prepared to (1) upload any supplementary materials (e.g., audio samples, code, supplementary data) – again, preserving author anonymity; (2) write one line of text summarizing the main take-home message of their work; and (3) indicate any reviewers who present a conflict of interest and should not review the work (this typically includes any current or recent mentors, mentees, colleagues, or collaborators). Review decisions will be sent by July 9, 2021 , and camera-ready versions of accepted papers will be due by August 2, 2021.
Special Call for Papers: Cultural Diversity in MIR
For ISMIR2021, there is also a special Call for Papers on Cultural Diversity in MIR. As described on the Call for Papers web page, any paper which meets the ISMIR paper guidelines and addresses an MIR topic of interest "with a focus on non-Western music or cross-cultural studies" may be submitted in this track. These papers will be reviewed by carefully chosen reviewers; accepted papers will have the same presentation format as other papers, but will receive special recognition on the conference website and be eligible for a special award in the themed track. This is an optional submission track, and any paper may still be submitted in the regular track.
The ISMIR paper reviewing process
ISMIR papers undergo a one-stage, two-tier reviewing process. By one stage, we mean that submissions are accepted or rejected in a single round of review ; there is no rebuttal process or option to revise and resubmit. Consequently, accepted papers are typically those requiring no more than minor revisions. The two-tier reviewing process involves a pool of regular reviewers plus a meta-reviewer assigned to each paper. Once the initial reviews are completed, the meta-reviewer coordinates a discussion phase with the reviewers and synthesizes the reviews into a final recommendation to the Scientific Program Chairs. Every paper will be evaluated by at least three reviewers.
ISMIR paper reviews are double blind , meaning that the identities of the authors are unknown to the reviewers, and vice versa. Therefore, it is critical that authors not identify themselves in any way in the initial submission. Authors of accepted papers have a window of time after acceptance to prepare the "camera-ready" version of their paper (i.e., the version that appears in the proceedings), at which time they will de-anonymize the submission.
ISMIR has an established set of criteria on which submitted papers are evaluated: Novelty, scholarly/scientific quality, reusable insights, thought-provoking potential, appropriateness of topic, importance, and presentation/readability (see the Call for Papers for more information).
To make these concepts more concrete, we strongly encourage every ISMIR author to closely examine ISMIR paper reviewer materials. The ISMIR Society maintains an online webpage of Reviewer Guidelines as well as a link to a document containing example ISMIR reviews compiled by MIR researcher Tom Collins. Moreover, an example reviewer form can be accessed on page 3 of the example review document. Authors can and should evaluate their paper against these metrics prior to submission in order to maximize the success of their submission.
Authors are encouraged to keep in mind some common pitfalls that cause ISMIR submissions to be rejected:
- The work is unfinished. As mentioned previously, the scientific work must be completed in order for a submission to be accepted. In general, submissions that propose analyses but do not perform them, or state that analyses will be completed in time for the camera-ready submission deadline, will not be accepted.
- The relevance of the submission to MIR is not clear. This arises particularly for submissions that do not fall into the "core" topics of MIR. For example, if an author has conducted a study that does not have direct MIR applicability (e.g., in psychoacoustics, mathematics, information science), they are encouraged to devote some effort to framing the study's potential relevance for MIR applications.
- The work is not sufficiently novel. This comment indicates that no significant new contribution(s) are evident in the presented work when compared to previously published work on a similar problem or topic.
- Quantitative analyses lack evaluation metrics or statistics. Submissions may be negatively reviewed if authors have not supported their claimed outcomes with sufficient rigor. Reporting of quantitative results should be accompanied by appropriate statistical analyses. When appropriate, authors are encouraged to use available objective evaluation methods (such as those described in https://craffel.github.io/mir_eval/) as much as possible and also ensure that the same evaluation criteria are applied across systems that are being compared in their paper. Other good practices include dataset partitioning in a manner that avoids train-test leakage in machine learning-based work. We also recommended following Reproducible Research guidelines as much as possible (https://ismir.net/resources/reproducible/).
- The results are not generalizable or do not consider a sufficient sample size. For instance, machine learning experiments must involve either standard data sets or other data sets that are of sufficient size and are clearly described in the paper. It is important to present system or implementation details that make for reproducible research. Similarly for research involving human subjects, authors should be cognizant of standard sample sizes for survey, interview, and experimental research by consulting published literature in the appropriate domain. For qualitative studies, the goal may not be to generalize the results to a wider population but rather, to gain deeper insights into a problem; in these settings, the expected sample size tends to be much smaller. It can be helpful to clearly make this point when describing the study design and/or limitations.
- Please keep in mind that studies with qualitative data may have a different research goal (such as gaining deeper insights into a problem) rather than generalizing the results to a wider population, and also the expected sample size differs from quantitative studies.
- The submission has major formatting errors. Authors may not change the formatting specifications of the provided paper templates. In addition, submissions which have incomplete metadata, are not anonymized, do not use one of the provided templates, or do not adhere to the page limits will be automatically rejected.
Not all papers will be accepted; the acceptance rate for ISMIR conference papers tends to be between 40-50%.
We recognize that a paper rejection is a disappointing outcome of hard work on the part of the authors. We would like to highlight some options within and beyond the ISMIR conference that authors can consider as a future home for their work:
- Submit to this year's ISMIR conference in a different format. For instance, creative works may be suitable music submissions; and submissions that are smaller in scope, or whose authors are new to ISMIR or to the full-paper format, may be a good fit with the late-breaking demo format.
- Resubmit to next year's ISMIR conference. This is not uncommon for papers receiving suggested major revisions, or papers that did not contain completed work in their initial submission.
- Submit a longer-format paper to TISMIR. Some research endeavors are too large in scope to be adequately reported in the ISMIR conference page limit, and receive negative reviews for exceeding the page limit or for providing insufficient detail in the literature review, methodology, or results sections. Authors who receive these types of reviews may consider submitting to the Transactions of the International Society for Music Information Retrieval (TISMIR), which is the journal of the ISMIR Society and complements the ISMIR conference proceedings.
Frequently asked questions (FAQs)
I need an additional week to work on my paper submission. Can I get an extension?
In order to adhere to the review and notification timeline, there will be no extension to the deadlines listed in the Call for Papers (abstracts due May 8; full papers due May 15; AoE).
I can't fit all of my content in the page limit. Is it okay to have 1-2 pages of extra content?
Unfortunately there is no option to submit longer papers to the ISMIR conference, and papers that do not adhere to the 6+n format will be rejected. If you find your submission is too long for the ISMIR page limits, it may be a better fit for the Transactions of the International Society for Music Information Retrieval (TISMIR) journal.
My ISMIR submission is already uploaded to a preprint server. Will this cause my submission to be automatically rejected?
Authors are strongly encouraged not to post submitted ISMIR papers to any online venue during the review process. Doing so may interfere with the double-blind review process and may cause your paper to be rejected.
I am interested in submitting an ISMIR paper, but I work in a different field from MIR. How do I know if my research would be appropriate for the ISMIR conference?
We expect the research to broadly fit in the list of topics available in the Call for Papers. You are encouraged to look up past ISMIR proceedings to better understand the scope of the conference (see: https://www.ismir.net/conferences/), and to clearly state in the paper how the work is directly or potentially relevant to MIR.
I am interested in submitting an ISMIR paper but am not sure whether I have funds to register for the conference if my paper is accepted. What should I do?
The ISMIR2021 organizers are committed to making the conference accessible and will be offering a number of fee reductions and grants. Therefore, authors are encouraged to submit any paper that meets the guidelines of the Call for Papers.
Do you have additional questions about ISMIR paper submissions that are not addressed here? Email the ISMIR2021 Scientific Program Chairs at firstname.lastname@example.org.