Hardcoded question support

.gitignore

@@ -125,4 +125,5 @@ settings.yaml
 nimbus_api.iml
 
 # models
-*pkl
+*pkl
+id_rsa

QA.py

@@ -49,6 +49,8 @@ def __init__(self, q_format, db_query, format_answer, db):
         self.format_answer = format_answer
 
     def answer(self, extracted_vars):
+        if self.db_query is None:
+            return self.format_answer({"normalized entity": ""}, dict())
         db_data = self.db_query(extracted_vars, self.db)
         answer = self.format_answer(extracted_vars, db_data)
         return answer
@@ -239,6 +241,8 @@ def _chain_db_access(
 
 # Actually returns partial[Dict[str, Any]]
 def chain_db_access(fns: List[DB_Query]) -> DB_Query:
+    if fns is None:
+        return None
     return functools.partial(_chain_db_access, fns)
 
 
@@ -287,6 +291,9 @@ def generate_qa_pairs(qa_pairs: Tuple[str, str], db: NimbusMySQLAlchemy):
                     db_access_fns.append(get_property_list(prop, joiner, table))
                     tokens[i] = "{db_" + prop + "}"
 
+        if not db_access_fns:
+            db_access_fns = None
+
         o = QA(
             q_format=q,
             db_query=chain_db_access(db_access_fns),

nimbus.py

@@ -28,12 +28,14 @@ def __init__(self, db: NimbusMySQLAlchemy):
             self.classifier.train_model()
             self.classifier.load_latest_classifier()
 
-
     def answer_question(self, question):
         ans_dict = self.predict_question(question)
         print(ans_dict)
         try:
-            qa = self.qa_dict[ans_dict["question class"]]
+            if ans_dict is None:
+                qa = self.qa_dict[self.classifier.classify_question(question)]
+            else:
+                qa = self.qa_dict[ans_dict["question class"]]
         except KeyError:
             # Printed if question isn't found. This occurs because the training set is broader
             # than the answerable question set.
@@ -64,6 +66,9 @@ def predict_question(self, question):
         # Get dictionary of extracted variables + info from question
         nlp_props = self.variable_extractor.extract_variables(question)
 
+        if nlp_props is None:
+            return None
+
         # Add classified question to nlp_props dictionary
         nlp_props["question class"] = self.classifier.classify_question(
             nlp_props["normalized question"]

nimbus_nlp/models/features/overall_features.json

@@ -1 +1 @@
-{"prereqs": 0, "COURSE": 0, "be": 0, "course": 0, "require": 0, "take": 0, "unit": 0, "count": 0, "typically": 0, "offer": 0, "quarter": 0, "go": 0, "4": 0, "class": 0, "term": 0, "prerequisite": 0, "have": 0, "recommend": 0, "need": 0, "order": 0, "topic": 0, "cover": 0, "title": 0, "call": 0, "code": 0, "time": 0, "year": 0, "available": 0, "fulfill": 0, "preq": 0, "soon": 0, "description": 0, "learn": 0, "credit": 0, "worth": 0, "number": 0, "corequisite": 0, "concurrent": 0, "enrollement": 0, "list": 0, "lab": 0, "PROF": 0, "office": 0, "hour": 0, "right": 0, "later": 0, "today": 0, "tomorrow": 0, "email": 0, "phone": 0, "contact": 0, "information": 0, "reach": 0, "room": 0, "in": 0, "address": 0, "visit": 0, "host": 0, "job": 0, "lecturer": 0, "professor": 0, "position": 0, "Cal": 0, "Poly": 0, "'s": 0, "place": 0, "find": 0, "campus": 0, "building": 0, "day": 0, "CLUB": 0, "club": 0, "box": 0, "college": 0, "affiliate": 0, "President": 0, "Advisor": 0, "do": 0, "brief": 0, "overview": 0, "give": 0, "chief": 0, "touch": 0, "get": 0, "general": 0, "adviser": 0, "usually": 0, "currently": 0, "advisor": 0, "teaching": 0, "week": 0, "platform": 0, "virtual": 0, "zoom": 0, "vroom": 0, "Zoom": 0, "Skype": 0, "see": 0, "section": 0, "teach": 0, "type": 0, "department": 0, "found": 0, "tell": 0, "student": 0, "attend": 0, "faculty": 0, "ratio": 0, "big": 0, "acre": 0, "program": 0, "bacherlor": 0, "degree": 0, "master": 0, "minor": 0, "credential": 0, "make": 0, "create": 0, "help": 0, "ask": 0, "kind": 0, "question": 0, "69": 0, "not related": 0}
+{"prereqs": 0, "COURSE": 0, "be": 0, "course": 0, "require": 0, "take": 0, "unit": 0, "count": 0, "typically": 0, "offer": 0, "quarter": 0, "go": 0, "4": 0, "class": 0, "term": 0, "prerequisite": 0, "have": 0, "recommend": 0, "need": 0, "order": 0, "topic": 0, "cover": 0, "title": 0, "call": 0, "code": 0, "time": 0, "year": 0, "available": 0, "fulfill": 0, "preq": 0, "soon": 0, "description": 0, "learn": 0, "credit": 0, "worth": 0, "number": 0, "corequisite": 0, "concurrent": 0, "enrollement": 0, "list": 0, "lab": 0, "PROF": 0, "office": 0, "hour": 0, "right": 0, "later": 0, "today": 0, "tomorrow": 0, "email": 0, "phone": 0, "contact": 0, "information": 0, "reach": 0, "room": 0, "in": 0, "address": 0, "visit": 0, "host": 0, "job": 0, "lecturer": 0, "professor": 0, "position": 0, "Cal": 0, "Poly": 0, "'s": 0, "place": 0, "find": 0, "campus": 0, "building": 0, "day": 0, "CLUB": 0, "club": 0, "box": 0, "college": 0, "affiliate": 0, "President": 0, "Advisor": 0, "do": 0, "brief": 0, "overview": 0, "give": 0, "chief": 0, "touch": 0, "get": 0, "general": 0, "adviser": 0, "usually": 0, "currently": 0, "advisor": 0, "teaching": 0, "week": 0, "platform": 0, "virtual": 0, "zoom": 0, "vroom": 0, "Zoom": 0, "Skype": 0, "see": 0, "section": 0, "teach": 0, "type": 0, "department": 0, "found": 0, "tell": 0, "student": 0, "attend": 0, "faculty": 0, "ratio": 0, "big": 0, "acre": 0, "program": 0, "bacherlor": 0, "degree": 0, "master": 0, "minor": 0, "credential": 0, "make": 0, "create": 0, "help": 0, "ask": 0, "kind": 0, "question": 0, "not related": 0}

nimbus_nlp/variable_extractor.py

@@ -70,7 +70,8 @@ def extract_variables(self, sent):
 
         # Make the prediction
         request = self.get_prediction(sent)
-
+        if not request.payload:
+            return None
         # Obtain the entity in the sentence
         entity = request.payload[0].text_extraction.text_segment.content