Prepare your data

Before depositing your data in DataverseNO (including the different sub-archives, e.g. UiT Open Research Data, TROLLing, etc.) you have to make sure your dataset(s) comply with our guidelines below. In brief, good practice for preparing research data for archiving may be summarized as follows:

  • Use consistent and comprehensible file names.
  • Add your data in a persistent file format in addition to the original file(s).
  • Describe your data in (a) ReadMe file(s).

For more detailed guidelines, see below:

1 File naming

Following good practice for file naming and organizing makes it much easier to find the right data file, not just for you, but also for your collaborators, and later on for other researchers who may re-use your data. Please make sure your file names comply with the following fundamental file naming recommendations:

  • Files must be named consistently.
  • File names must be descriptive, but short (< 25 characters).
  • Do not use spaces. Instead, use underscores (e.g. first_study), hyphens (e.g. first-study) or camel case (FirstStudy).
  • Avoid characters like “ / \ : * . ? ‘ < > [ ] ( ) & $ æÆ øØ åÅ äÄ öÖ …
  • Use the international dating convention YYYY-MM-DD (e.g. 2017-10-25).

2 Persistent file formats

What are persistent file formats?

The choice of a persistent file format is crucial in order to ensure that your data will be readable also in the future. Some file formats are more likely to allow long-term readability than others are. Such formats are usually

  • non-proprietary
  • open, with documented international standards
  • in common usage by the research community
  • using standard character encoding, preferably Unicode (e.g. UTF-8)
  • uncompressed (space permitting)

The table below gives an overview of persistent vs. non-persistent file formats for a selection of document types. [1] When uploading your data to the archive, please make sure you add your files in a persistent format in addition to the original file format. Make also sure that all of your files contain a valid file extension, e.g. .txt, .pdf.

File type Persistent file formats (examples) Non-persistent file formats (examples)
  • Uncompressed and lossless Wav or AIFF (.wav/.aiff)
  • Compressed and lossless FLAC (.flac)
  • Compressed and lossy Mp3 (.mp3)
  • AAC (.m4a)
  • Monkey’s Audio (.ape)
  • Ogg Vorbis (.ogg)
  • Windows Media Audio (.wma)
Container files Container files are not recommended. If necessary, us the following formats:

    • zip
    • tar

Note! Container files are automatically unpacked when uploaded, and must therefore be packed twice. That way, the inner container will be preserved.

  • 7z
  • gz
  • rar
  • Uncompressed TIFF (.tif or .tiff)
  • Compressed and lossless PNG (.png)
  • Compressed and lossy JPEG (.jpg)
  • Adobe Photoshop (.psd)
  • Apple Picture File (.pct)
  • Graphics Interchange Format (.gif)
  • Raw Image Data File (.raw)
  • Windows Bitmap (.bmp)
Text (slides, illustrations)
  • PDF/A (.pdf) combined with original file
  • PowerPoint (.pptx)
Text (tables)
  • Tab separated Unicode plain text (.txt)
  • Excel (.xlsx)
Text (text)
  • Plain text (.txt)

If formatting needed:

  • XML, PDF/A (.pdf) combined with original file
  • Word (.docx)
  • HTML
Transcription File format:

  • PDF/A (.pdf) combined with original file
  • PDF/A (.pdf) combined with Comma/Tab Separated Values (.csv/.txt)


  • Unicode IPA (e.g. Charis SIL, Doulos SIL, Gentium Plus, Andika), ASCII SAMPA
File format:

  • Word (.doxc)
  • Excel (.xlsx)


  • Transcription legacy fonts (SIL IPA(93))
  • MPEG-4 (.mp4)
  • AVI (.avi)
  • Flash Video (FLV)
  • Quicktime (.mov)
  • Windows Media Video (WMV)
Workspace dump for Matlab, R, S-Plus, SPSS or similar Include:

  • Basic data as tab separated Unicode plain text (.txt)
  • Script(s) as Unicode plain text (.txt)
  • The different workspace dump formats, e.g. .mat, RData, .R
Work space dump formats for mass spectrometry
  • mzML (.mzML)[2]
  • Agilent D (.D)
  • Bruker BAF (.BAF)
  • Bruker FID (.FID)
  • Chromtech DAT (.DAT)

[1] The list of file formats in the column “Non-persistent file formats” is non-exhaustive and includes the formats considered the ones used most commonly. If your dataset contains file formats not listed here, please contact the support services of your home institution.
[2] Read more about this format here.

How to save or convert your data into a consistent format?

This section contains information on the following document types: Audio, container, image, text, transcription, and video. If your data contain types not listed here, please contact the support services of your home institution.


  • Recording:
    The quality of your audio file depends on the purpose of your recording. If the recording is of such nature that acoustic details are irrelevant, the mp3 format is sufficient. Note however, that mp3 is a lossy compression format: Information in the speech signal is irreversibly discarded during recording and can therefore be considered less suited for speech analysis in the case of data reuse.Given that the mp3-format reduces the reusability of your data, we advise recording in an uncompressed format, .wav or .aiff.
  • Conversion:
    If space is an issue, you can convert the uncompressed .wav and .aiff-files after recording. We recommend a format that does not remove information, like FLAC (Free Lossless Audio Codec). Conversion to FLAC is fully reversible, i.e. the original sound file is restored when decompressed.File conversion can easily be done in free software like Audacity ( or Praat (

Container files

We do not recommend to use container files. By default, ZIP container files containing up to 1 000 files will be automatically unpacked when uploaded to the DataverseNO. If you want to retain the original folder structure, you have to tag the files with the respective folder names. If you for some reasons have to use container files, please follow the recommendations below:

  • Use container files with extensions .zip or .tar (do not use .7z, tar.gz, .rar, and so on). The tar format is preferred for long-term archiving because it is openly-documented.
  • Use one of the following tools to pack your files into a container:
    • 7-Zip (for Windows)
    • Keka (for Mac, or use function tar on command line)
  • Do not use compression or encryption when packing your files into containers.


  • Compression:
    Images are often compressed to reduce the amount of redundant or irrelevant data information. This does not mean that the quality reduction is visible to the human eye. For instance, PNG-files maintain all information in the image. As for JPEG-files – a widely used file format – the rate of compression can be manipulated: Depending on type of image and potential size issues, you must, in each case, determine how much compression is advisable, with regard to both reuse and sharing of your image files.
  • Conversion:
    If your images are stored in a format considered non-persistent (see the section What are persistent file formats? above), they must be converted to JPEG, PNG or TIFF. Conversion can easily be done in the software Paint (Windows), Preview (Mac) or GIMP Image Editor (Linux). There are numerous free image converters. However, before using one of these, it is advisable to read any terms of use.


Plain text

If your data is represented in plain text, requiring little or no formatting, you are recommended to create and save your data as plain text files (.txt). You may use a simple text editor, e.g. gedit, TextEdit or WordPad. If you use a more advanced text editor when structuring your data, e.g. Microsoft Word or LibreOffice Writer, you must still save it in plain text format. To do so, select “Save as file type: Plain text (.txt)” in the menu File > Save As.

Formatted text

If your data contains formatted text, e.g. including essential line breaks, tabs, figures, we recommend you to convert your data file into a PDF/A file (.pdf). The original text file as well as the PDF/A file must be uploaded. The same procedure must be carried out if you use a text editor like Microsoft Word or LibreOffice Writer when structuring your data, or a presentation editor like Microsoft PowerPoint or LibreOffice Impress.

To create a PDF/A file in Microsoft Word:
Mac (2011): Print > PDF > Save as Adobe PDF > Adobe PDF Settings: PDF/A-1b: 2005 (CMYK). Note that this option requires Adobe Acrobat. If this is not available, save the file as plain PDF.

Windows (2013): Save as Adobe PDF > File type: PDF files > Options: Create PDF/A-1a: 2005 compatible file

To create a PDF/A file in LibreOffice Writer:
Linux: Save as PDF > Check the PDF/A-1a box > Export.

To save a PDF file as a PDF/A file in Adobe Acrobat (Pro or similar):
Save As Other > More Options  > PDF/A.

Tabular text

Tabular text data must be provided as Unicode-encoded text files (.csv/.txt). If you have stored your data in a spreadsheet software like Microsoft Excel or LibreOffice Calc, the following instructions show you how to convert it to a recommended format:

Microsoft Excel (Mac, Windows):

  • (On a laptop: Click More options below the file type field displaying Excel Workbook (*.xslx))
  • Choose File > Save as > Choose folder
  • In the option Save as type, choose Text (Tab delimited) (*.txt)
    (Note! Do not choose Unicode Text (*.txt))
  • In Tools, choose Web options
  • Choose the tab Encoding
  • In the field Save this document as, choose Unicode (UTF-8), and then click OK
  • Choose the tab Fonts
  • In the Character set window, choose Multilingual/Unicode/Other script, and click OK
  • Click Save
  • Confirm by clicking Yes
  • Note: This process has to be repeated for each sheet in the Excel workbook

LibreOffice Calc (Linux, Mac, Windows):

  • Click File > Save As
  • For each sheet in the LibreOffice Calc workbook, proceed as follows:
    • Linux and Windows: In the data export dialogue window, select
      • Text encoding/Character set: Unicode (UTF-8)
      • Field delimiter: {Tabulator} (= recommended)
      • Text delimiter: none (erase the prefilled one from the field)
    • Mac: In the field File type, select “Text CSV (.csv)”. In the data export dialogue window, select
      • Character set: Unicode (UTF-8)
      • Field delimiter: {Tab}
      • Text delimiter: “ (double quotation mark)

If the very graphical layout of your tabular data is essential in order to understand them, you must also upload a PDF/A version of the document. Also, if your tabular text data contain figures, charts or other kinds of graphical elements that are essential for understanding your data, it is recommended that you convert these elements into PDF/A documents. See conversion procedure for formatted text above.


  • Font:
    All transcriptions must be made using Unicode-encoded fonts, e.g. IPA Doulos SIL.[1] For phonetic transcriptions, SAMPA (Speech Assessment Methods Phonetic Alphabet, ASCII characters)[2] is an alternative to IPA. If the recommended font is not available for the type of transcription your dataset requires, it is imperative to include a separate ReadMe file in your dataset with instructions about how to read the transcriptions.[3] Note that the font package itself must not be uploaded, given copyright restrictions.
    [1] To download SIL Fonts, cf.
    [2] For an overview of SAMPA symbols, cf.
    [3] Cf. for instance an example in the file “To read the Church Slavonic transcriptions.pdf” in Eckhoff (2015), cf.
  • Conversion:
    If your videos are stored in a format considered unacceptable (see the section What are persistent file formats? above), these must be converted to the MPEG-4 format. If you do not have license to any professional conversion software, we advise you to use the VLC Media Player (standard application on both Mac and Windows), or an online free image converter. However, before using any free converter, it is advisable to read any terms of use.


  • Matlab, R, S-Plus, SPSS or similar:
    Most softwares for statistical analysis allow you to save the basic data as (or export them to) a plain text format (.txt). In addition, you must copy the script, and save it as plain text in a text editor.
  • Software for mass spectrometry:
    Guidelines on how to convert .mid files to .mzML can be found here. If you are unfamiliar with the command line in Windows, please contact user support at your home institution.

3 How to describe your data

In order for users to be able to understand and reuse your data, it is essential that you describe it in a comprehensible and consistent manner. Data come in many different forms, and for most types, there is no common standard of description. In this section, we present guidelines on how you must prepare and describe data for archiving in DataverseNO.

Your data description must be provided in a file named “ReadMe” together with your data files. You must save your ReadMe file(s) as a Unicode UTF-8 plain text file (.txt). In case you need to use illustrations or special characters, you may save your ReadMe file(s) as PDF/A (see the section on persistent file formats above for more information about these file formats). Use forced numbering in the ReadMe filename (e.g. “00_ReadMe.txt”) to make it appear on the top of the file overview.

First in your ReadMe file(s) you must give an overview and short description of the files contained in your dataset. The remaining contents of your ReadMe file(s) will vary according to what kind of data you are going to archive. Below we give some recommendations for ReadMe files for two common types of data, tabular data and computer scripts.

Tabular data

It is advisable to upload a separate ReadMe file with a comprehensive description of the data file, including the data in each column, the data format and the standard(s) used. This can additionally be inserted into the Description field in the Citation Metadata tab.

  • Columns and column headings:
    For each column in your tabular text file (.csv or .txt; see above) you must indicate what kind of data it contains, and what data format the values have. Column headings must be meaningful and not too long. Make sure you do not use duplicate column headings within a file. Use only alphanumeric characters, underscores, or hyphens in column headings. It is good practice to have column headings start with a letter. If possible, indicate units of measurement in the column headings.Use only the first row for column headings, otherwise rows may be missed when your data is imported to spreadsheet software or other utilities. Example of good column headings: vowel_length_ms, record_time, language_name, pos.
  • Data values and formatting:
    Use standard codes or names when possible (e.g. ISO code for language names) and established tag sets for POS/parts of speech (e.g. CLAWS2 Tagset). Avoid using special characters, such as commas, semicolons, or tabs, in the data itself. This might cause trouble when the data file is imported into a spreadsheet, or read by other software. If such characters are nevertheless necessary in the presentation of your data, please specify their use in the ReadMe file.
  • Examples of tabular data description:
    – The column “vowel_length_ms” contains values for the vowel length in milliseconds of the analyzed items in the dataset. Only integer numbers are used, e.g. 45, 32, 11.
    – The column “record_time” contains values for the time when the record was made. The time format used is YYYY-MM-DD hh:mm, e.g. 2014-03-15 17:21.
    – The column “lang_name” contains values for the name of the analyzed languages. The ISO 639-2 Code format is applied:
    dan       Danish
    nob      Norwegian Bokmål
    swe      Swedish

    – The column “pos” contains values for the part of speech of the analyzed items. The applied tag set is the CLAWS2 Tagset:
    NP        proper noun, neutral for number (Indies, Andes)
    NP1      singular proper noun (London, Jane, Frederick)
    NP2      plural proper noun (Browns, Reagans, Koreas)

Source code/script

Another common data type are scripts used in statistical analysis. Before archiving, make sure you add a description for each step used in the script. Below, we present an example, taken from TROLLing[1]:

[1] Janda et al. (2014), cf.

4 File size

The size of each individual file upload must not exceed 8 Gb. If you want to upload files that are larger than 8 Gb in total, you have to upload them in several uploads. You do this by saving the dataset after each upload. As of today, there is no upper limit to the size of a dataset, but we recommend that you contact the support services of your home institution if you wish to add a dataset with a total file size of more than 50 Gb.

5 References

Parts of the guidelines above have been adapted from several sources, including

Data Management General Guidance. Curation Center of the California Digital Library, University of California.

Praat beginners’ manual by Sidney Wood.

Preparing tabular data for description and archiving. Research Data Management Group, Cornell University.

Recommendations for uploading data. ETH-Bibliothek.

Sustainable Formats and Conversion Strategies at the Bentley Historical Library. Version 1.0, November 9th, 2011.

For questions, comments or suggestions, see our support page.