Update auditqa/retriever.py

auditqa/retriever.py

@@ -9,20 +9,20 @@ model_config = getconfig("model_params.cfg")
 def create_filter(reports:list = [],sources:str =None,
                   subtype:str =None,year:str =None):
     if len(reports) == 0:
-        print("defining filter for sources:{},subtype:{},year:{}".format(sources,subtype,year))
+        print("defining filter for sources:{},subtype:{},year:{}".format(sources,subtype))
         filter=rest.Filter(
                 must=[rest.FieldCondition(
                         key="metadata.source",
                         match=rest.MatchValue(value=sources)
                     ),
                     rest.FieldCondition(
-                        key="metadata.subtype",
+                        key="metadata.filename",
                         match=rest.MatchValue(value=subtype)
                     ),
-                    rest.FieldCondition(
-                        key="metadata.year",
-                        match=rest.MatchAny(any=year)
-                    ),])
+                    #rest.FieldCondition(
+                    #    key="metadata.year",
+                    #    match=rest.MatchAny(any=year)
+                    ])
     else:
         print("defining filter for allreports:",reports)
         filter=rest.Filter(
@@ -35,13 +35,13 @@ def create_filter(reports:list = [],sources:str =None,
     return filter
 
 
-def get_context(vectorstore,query,reports,sources,subtype,year):
+def get_context(vectorstore,query,reports,sources,subtype):
     # create metadata filter
-    filter = create_filter(reports=reports,sources=sources,subtype=subtype,year=year)
+    filter = create_filter(reports=reports,sources=sources,subtype=subtype)
 
     # getting context
     retriever = vectorstore.as_retriever(search_type="similarity_score_threshold", 
-                                         search_kwargs={"score_threshold": 0.6, 
+                                         search_kwargs={"score_threshold": 0.4, 
                                                         "k": int(model_config.get('retriever','TOP_K')), 
                                                         "filter":filter})
     # re-ranking the retrieved results