h3abionet / afpo

AfPO: African Population Ontology
2 stars 0 forks source link

Population_group inconsistency violation #24

Closed anitacaron closed 7 months ago

anitacaron commented 8 months ago

Creates a QC for consistency in population group and country of origin.

Fixes #25

anitacaron commented 7 months ago

@daniwelter @Melek-C @abenkahla I've added the missing population_group in the country, but it got removed again because of the reasoning and reduce process (remove any redundant logical axiom).

In the case of Berber Northern, its three children are already listed as a population group in Algeria, so logically, Berber Northern is listed as part of the Algerian population group. I'll show Protégé screenshots to explain better the situation.

Algeria without population_group Berber Northern Screenshot 2024-03-15 at 11 56 28

Query to list population_group for Berber Northern Screenshot 2024-03-15 at 11 58 54

Explanation Algeria is on the list Screenshot 2024-03-15 at 11 59 05

The same happens for Lybia. If I remove in Lybia population_group Berber Northern Screenshot 2024-03-15 at 13 40 13

It's inferred by its two children: Screenshot 2024-03-15 at 13 15 38

However, it fixed for some cases, for example, Egypt which is the only assertion. Screenshot 2024-03-15 at 13 15 53

daniwelter commented 7 months ago

@anitacaron in the case where the countries can be inferred, such as the Berber Northern example, I think this is fine. It's only countries where the reverse link cannot be inferred that I consider a problem.

Melek-C commented 7 months ago

Hi @anitacaron, after checking the last version I found that the information exist in both as in the screenshot and also true for the other countries.

image

anitacaron commented 7 months ago

@Melek-C Yes, I added the missing axioms in the afro-edit.ofn, but because I'm using afpo-full.owl, which is the afpo-edit.ofn, after reasoning and removing duplicated axioms, the QC is still falling. I'll change the afpo-full.owl to materialize the inferences only for QC purposes.

Melek-C commented 7 months ago

Ok @anitacaron, it should be ok then?

anitacaron commented 7 months ago

@Melek-C I didn't find more issues related to population_group, but I tested the other way, population without the country of origin, and found some inconsistencies.

population population_label country country_label Comment Is it right?
AfPO0000289 Kanembu dbpedia:Cameroon Cameroon Inferred by subClassOf Kanuri ABK&MC:No
AfPO0000288 Kwatay dbpedia:Cameroon Cameroon Kwatay hasCountryOfOrigin Senegal. Missing hasCountryOfOrigin Cameroon. ABK&MC:Yes for Senegal AND No for Cameroon
AfPO0000080 Masalit dbpedia:Cameroon Cameroon Masalit hasCountryOfOrigin Chad and Sudan. Missing hasCountryOrigin Cameroon. ABK&MC:Yes for Chad and Sudan AND No for Cameroon
AfPO0000543 Dan dbpedia:Ivory_Coast Côte d'Ivoire Inferred by subClassOf Mande Western ABK&MC: hasCountryOfOrigin 'Côte d'Ivoire'
AfPO0000544 Kabye dbpedia:Guinea Guinea Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Guinea. ABK&MC: Kabye hasCountryOfOrigin Togo.
AfPO0000347 Pedi dbpedia:Lesotho Lesotho Inferred by subClassOf Sotho ABK&MC: No
AfPO0000544 Kabye dbpedia:Liberia Liberia Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Liberia. ABK&MC: No Kabye hasCountryOfOrigin Togo only
AfPO0000342 Shona dbpedia:Mozambique Mozambique Inferred by subClassOf Bantu Eastern ABK&MC: No.
AfPO0000074 Bambara dbpedia:Senegal Senegal Bambara hasCountryOfOrigin Côte d'Ivoire, The Gambia and Mali. Missing hasCountryOfOrigin Senegal. ABK&MC: Yes please add Senegal as hasCountryOfOrigin
AfPO0000543 Dan dbpedia:Senegal Senegal Dan hasCountryOfOrigin The Gambia and Liberia. Missing hasCountryOfOrigin Senegal. ABK&MC: Yes
AfPO0000161 Dyula dbpedia:Senegal Senegal Dyula hasCountryOfOrigin Burkina Faso, Côte d'Ivoire, Ghana and Mali. Missing hasCountryOfOrigin Senegal. ABK&MC: Do not add Senegal as hasCountryOfOrigin
AfPO0000544 Kabye dbpedia:Sierra_Leone Sierra Leone Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Sierra Leone. ABK&MC: No Kabye hasCountryOfOrigin Togo only
AfPO0000338 Lemba dbpedia:South_Africa South Africa Inferred by subClassOf Bantu Southern ABK&MC: 'South Africa' should be hasCountryOfOrigin
AfPO0000084 Bulu dbpedia:São Tomé and Príncipe São Tomé and Príncipe Inferred by subClassOf Beti-Pahuin ABK&MC: No
AfPO0000572 Ewondo dbpedia:São Tomé and Príncipe São Tomé and Príncipe Inferred by subClassOf Beti-Pahuin ABK&MC: No
AfPO0000336 Himba dbpedia:Tanzania Tanzania Himba using Population_group instead of hasCountryOfOrigin. Fixed. ABK&MC: perfect.
AfPO0000488 Kiga dbpedia:Tanzania Tanzania Inferred by subClassOf Bantu Eastern ABK&MC: Tanzania should be hasCountryOfOrigin
AfPO0000438 Tutsi dbpedia:Tanzania Tanzania Inferred by subClassOf Banyarwanda ABK&MC: Tanzania should be hasCountryOfOrigin
AfPO0000332 San dbpedia:Zimbabwe Zimbabwe San hasCountryOfOrgin South Africa, Botswana and Namibia. Missing hasCountryOfOrigin Zimbawbwe. ABK&MC: Yes, please add Zimbawbwe hasCountryOfOrigin

@Melek-C @abenkahla could you please check these cases and answer the last column?

anitacaron commented 7 months ago

For the Kabye cases. These are the remaining axioms that were not removed when we found Kabye and Kissi populations information were mixed at the Kabye population. I'll remove them.

population population_label country country_label Comment Is it right?
AfPO0000544 Kabye dbpedia:Guinea Guinea Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Guinea. No, remove
AfPO0000544 Kabye dbpedia:Liberia Liberia Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Liberia. No, remove
AfPO0000544 Kabye dbpedia:Sierra_Leone Sierra Leone Kabye hasCountryOfOrigin Benin and Togo. Missing hasCountryOfOrigin Sierra Leone. No, remove

Here are screenshots of the original file in WebProtege

Screenshot 2024-03-18 at 16 17 08 Screenshot 2024-03-18 at 16 17 00

abenkahla commented 7 months ago

Done.

anitacaron commented 7 months ago

Thank you, @abenkahla & @Melek-C! I'll make the changes in this PR.