Use case STM workflow

[az-ihsan]

Hi,

i have several questions regarding the STM workflows that i'm currently mapping to MDMC-NEP ontology.

1. Just to make sure Image Selection & Retrievement and Image Labelling Process are the way you get the data ready to be analyzed right which is here Structured& FAIR dataset?
2. About the Filtered Image, do you do some image filtering, machine learning, or deep learning to result in this image?
3. I'm still not quite understand what do you mean by Metadata Selection activity and why they used the Structure & FAIR Dataset and generated Filter Image. Could you explain it again to me maybe according to the workflow in the paper?

Hope @mpanighel and @EOsmenaj can answer this..

[EOsmenaj]

Hi,

1. 'Image selection & retrievement' can go under 'Data processing', instead 'Labelling process' under 'Data Analysis'.
2. Filtering is considered with respect to a result that was obtained by Machine learning techniques.
3. 'Metadata selection' is the action performed by a 'user' directly on the website: https://tridas.nffa.eu/dashboard_stm/
   Interacting with the platform, can find relevant images using just the selected metadata fields without providing specific information. Based on the number of fields selected, two plots are displayed: a quantiles plot to show the distribution of the images with respect to the values of the field and a scatter plot that shows the number of images for each combination of the two fields. The scatter plot allows the selection of a specific metadata combination to retrieve a new page containing a table with metadata fields for each image’s in that subset. On this page, researchers can select, filter, and search images based on their metadata values. The ID column consists of each image unique identifier in the database and, by clicking on it, the corresponding STM image is rendered and shown in a new page, where a download feature is included to obtain data, metadata, plot and provenance metadata for each image.

[mpanighel]

2. About the Filtered Image, do you do some image filtering, machine learning, or deep learning to result in this image?

More precisely I would say Image Labeling Process relies also on Machine Learning techniques and generates the Structured and FAIR dataset, in turn Metadata Selection includes some classical filtering on an image of the dataset (background subtraction and similar) ending up in the Filtered Image

[az-ihsan]

Yes, we discussed yesterday that Image Labelling Process could be a subclass/instance of Data Analysis

About the Metadata Selection and Filtered Image, there was a discussion on whether we change the name of both terms due to some confusion. Lets wait for @EOsmenaj to clarify this.

[EOsmenaj]

Hello,
concerning the change of the terms, we think that is better to take the ones we have since they are based on the specific work done and if we replace them with general ones such as data processing, for example, the provenance description will not be so useful in our case and we risk to lose specificity.
About the mapping: 'Image selection & retrievement' can be mapped with 'Data Processing'. 'Reference dataset' with 'Processed data', 'Image labeling process' with 'Data analysis lifecycle' since it includes data processing, data analysis, and data interpretation. 'Metadata selection' with 'Data processing' and 'Filtered image' with processed data. The 'filtered image' is the result of 2 actions: from one hand the research user applies filters to metadata to visualize the image and from the other hand the processing of that image from the visualization software.
Images can be downloaded together with the data files on them and the provenance information. Images and data files associated can be used for different purposes based on what a researcher is interested to do. The aim of the overall work was the curation of the original dataset (raw data) to make it almost FAIR compliant.

[az-ihsan]

Based on the discussion on 15.03 with @EOsmenaj @mpanighel @rossellaaversa @az-ihsan , we agreed to instantiate (make instances) this STM workflow to our MDMDC-NEP ontology.

[rossellaaversa]

structured and FAIR dataset could be mapped to the new provEntity:ProcessedDataOrAnalysedData: @EOsmenaj should check

[EOsmenaj]

Yes, I confirm.

[az-ihsan]

After the modularization, there is a need to revisit the mapping again.

[az-ihsan]

close by #68