added third kg generation script

general_working_directory/third_generation.py

@@ -0,0 +1,84 @@
+import base64
+import time
+from openai import OpenAI
+from owlapy.iri import IRI
+from owlapy.owl_ontology_manager import OntologyManager
+from owlapy.owl_property import OWLDataProperty
+from owlapy.owl_reasoner import OntologyReasoner, FastInstanceCheckerReasoner
+from rdflib import Graph, URIRef, Literal, RDFS, OWL, RDF
+from rdflib.namespace import XSD
+
+api_key = "token-tentris-upb"
+api_base = "http://tentris-ml.cs.upb.de:8501/v1"
+client = OpenAI(api_key=api_key, base_url=api_base)
+
+
+manager = OntologyManager()
+ontology = manager.load_ontology(IRI.create("file://fashionpedia-second-generation.owl"))
+base_reasoner = OntologyReasoner(ontology)
+reasoner = FastInstanceCheckerReasoner(base_reasoner=base_reasoner, ontology=ontology)
+g = Graph()
+g.parse("fashionpedia-second-generation.owl")
+
+has_llm_description = URIRef("http://example.org/hasLLMDescription")
+
+class_image = URIRef("http://example.org/Image")
+g.add((has_llm_description, RDF.type, OWL.DatatypeProperty))
+g.add((has_llm_description, RDFS.domain, class_image))
+g.add((has_llm_description, RDFS.range, XSD.string))
+
+def get_message(base64_image):
+    return client.chat.completions.create(
+    model="tentris",
+    messages=[{
+        "role": "user",
+        "content": [
+            {
+                "type": "text",
+                "text": "You are a fashion expert."
+                        "Your task is to give a short description of the apparel provided in the attached image."
+                        "You should focus only on the apparel presented to you. Don't describe the background."
+            },
+            {
+                "type": "image_url",
+                "image_url": {
+                    "url": f"data:image/jpeg;base64,{base64_image}"
+                }
+            }
+        ]
+    }],
+    temperature=0.1,
+    seed=1
+).choices[0].message.content
+
+
+def encode_image(image_path):
+    with open(image_path, "rb") as image_file:
+        return base64.b64encode(image_file.read()).decode('utf-8')
+
+
+def start_generation():
+    tasks = []
+    count = 0
+    for i in ontology.individuals_in_signature():
+        try:
+            image_ind = URIRef(i.str)
+            dprop1 = OWLDataProperty(IRI.create("http://example.org/hasFileName"))
+            image = "images/" + str(list(reasoner.data_property_values(i, dprop1))[0].get_literal())
+            base64_image = encode_image(image)
+            llm_description = get_message(base64_image)
+            g.add((image_ind, has_llm_description, Literal(llm_description, datatype=XSD.string)))
+            count += 1
+            print(f"{count:,}/45,623")
+        except Exception as e:
+            print(e)
+            g.serialize(destination="fashionpedia-third-generation.owl", format="xml")
+            exit(0)
+
+    g.serialize(destination="fashionpedia-third-generation.owl", format="xml")
+
+# Run the async loop
+if __name__ == "__main__":
+    start_time = time.time()
+    start_generation()
+    print(time.time() - start_time)