bonsamurais / exiobase-conversion-software Goto Github PK

See also BONSAMURAIS/rdf#15

Everything that will be deployed in the

http://rdf.bonsai.uno/data/

e.g.,

http://rdf.bonsai.uno/data/exiobase3_3_17/hsup

space needs to use the / scheme instead

This means that

 <http://rdf.bonsai.uno/data/exiobase3_3_17/huse#F_1111694>

needs to become

 <http://rdf.bonsai.uno/data/exiobase3_3_17/huse/F_1111694>

instead

Right now the readme is not clear on how toget the original data from
https://www.exiobase.eu/index.php/data-download/exiobase3hyb/

Each Flow is currently (and correctly) making use of the Unit of Measure Ontology with a hasUnit relation to a measure. The object of these triples is however not correct. Currently the object is a URI to a units-of-measure resource, example (http://www.ontology-of-units-of-measure.org/resource/om-2/tonne), but we have already defined the measures we use in the named graph unit.ttl, which we should instead use.

The project is dependent on the python package xlrd, as it is used by pandas to load xlsx files.

This dependency should be added to the requirements file.

Here column 4 in the CSV is not used ever, this is not a problem for the SUPPLY, but it is for the USE.

Also see example in
BONSAMURAIS/rdf#4

Input form RDF package group needed

The commit should be reverted

c7a82b4

and new code should be added to re-aggregate Flow quantities in the USE table

See discussion on:
#3

UPDATE: We keep the disagreggate values but a different process should produce a different dataset with aggregated values

Dear all,

Not quite sure if this is your preferred channel to discuss Agenda points. If not, please delete and let me know where to post.
Anyway, regarding the agenda point
"Write parsers for EXIOBASE format to RDF":
Happy to include a RDF export functionality to pymrio (https://github.com/konstantinstadler/pymrio). This would allow to generalize the BONSAI DB for any mrio.

Connected to that: I checked the tutorials for Leontief inversion and I see that you plan the EXIOBASE HIOT tables. Currently, I do not have a parser for the hybrid tables in pymrio, but if the format is stable I can work on that.
best
kst

The code should be refactored to follow the skeleton
https://github.com/BONSAMURAIS/python-skeleton

We were facing a challenge in detecting the determining product flow as theoretically these could be picked up from diagnol of the Supply tables. However, given that Exiobase is non- square i.e. 200 pdts x 164 activities, it is not that trivial to identify the determining flow directly.

Hence we have the following solution:

• We use a separate basic correspondence vector of determining products and the activities that helps us translate the information correctly into RDF

• Another method could be to square the Supply and Use tables before converting them to RDF. This was already done in the Mojo repository . Here the product flows are aggregated and the exclusive by-products are placed in a global market. This is useful to build the computational structure (A/Z matrix) but aggregating product flows also leads to loss of information.

Would like feedback

EXIOBASE has the following extensions:

• 39 resources, e.g. Blue water consumption, building stone, anthracite
• 5 land use types, e.g. arable land
• 66 emissions, e.g. NOx, Mirex, lead
• 66 emissions avoided by the use of manure
• 66 emissions from "unregistered waste", of which ~10 are non-zero
• 10 emissions from unregistered waste, drawn from same emissions list as above
• 8 "Other auxilliary account not elsewhere considered", e.g. "other non market feed", "use as feed of crop residues"

In addition, there are separate supply and use tables for 19 categories of waste/materials for the following:

• Total waste
• Packaging materials
• Materials in heterogeneous products
• Stock addition (only supply)
• "Supply of waste from materials accumulated previously in the society" (only supply)

Please provide your input on what is needed for a) a minimum viable product, and b) the final input of EXIOBASE into BONSAI. Each additional dataset represents a nontrivial burden to a group already stretched thin.