added chain of thought for event types and locations for improved acc…

…uracy

test.py

@@ -12,18 +12,21 @@
 from tracex.extraction.prototype import prompts as p
 from tracex.extraction.prototype import create_xes as x
 
-# text = open(u.input_path / "journey_synth_covid_0.txt").read()
-# df = ih.convert_text_to_bulletpoints(text)
-# print(df)
-# df = ih.add_start_dates(text, df)
-# df = ih.add_end_dates(text, df)
-# df = ih.add_durations(df)
-# # df = ih.add_event_types(df)
-# # df = ih.add_locations(df)
-
-# print(df)
-
-# ih.convert_dataframe_to_csv(df)
+text = open(u.input_path / "journey_synth_covid_0.txt").read()
+df = ih.convert_text_to_bulletpoints(text)
+print(df)
+df = ih.add_start_dates(text, df)
+print(df)
+df = ih.add_end_dates(text, df)
+print(df)
+df = ih.add_durations(df)
+print(df)
+df = ih.add_event_types(df)
+print(df)
+df = ih.add_locations(df)
+print(df)
+
+ih.convert_dataframe_to_csv(df)
 x.create_xes(u.output_path / "single_trace.csv", "test", "event_information")
 
 # output = 'The end date for the bulletpoint "Experiencing mild cough and fatigue" and the start date 20220108T0000 is 20230911T0000.'

tracex/extraction/prototype/prompts.py

@@ -133,7 +133,7 @@ def life_circumstances_prompt(sex):
     If there is only a month specified, use the first of this month as start date. If there is no date specified in the text conclude 'N/A'.
 """
 START_DATE_PROMPT = """
-    Here is the text and the bulletpoint for which you should extract the start date in the format YYYYMMDD with the postfix T0000! 
+    Here is the text and the bulletpoint for which you should extract the start date in the format YYYYMMDD with the postfix T0000!
     In case that you are not able to find a start date return the term "N/A". Only use the format YYYYMMDDTHHMM e.g. 20200401T0000!
     Explain step by step your conclusions if the date YYYYMMDDTHHMM is available or N/A.
 """
@@ -165,7 +165,7 @@ def life_circumstances_prompt(sex):
     Only return the date! Nothing else!
 """
 END_DATE_PROMPT = """
-    Here is the text and the bulletpoint with the start date for which you should extract the end date in the format YYYYMMDD with the postfix T0000! 
+    Here is the text and the bulletpoint with the start date for which you should extract the end date in the format YYYYMMDD with the postfix T0000!
     In case that you are not able to find a end date return the term "N/A". Only use the format YYYYMMDDTHHMM e.g. 20200401T0000!
     Explain step by step your conclusions if the end date YYYYMMDDTHHMM is available, if not calculate the average time of the activity and add this on the start date resulting as the end date.
 """
@@ -225,7 +225,8 @@ def life_circumstances_prompt(sex):
     The only output should be the event type!
 """
 EVENT_TYPE_PROMPT = """
-    Here is the bulletpoint for which you should extract the event type:
+    Here is the bulletpoint for which you should extract the event type.
+    Explain step by step your conclusions your choice of location: 'Symptom Onset', 'Symptom Offset', 'Diagnosis', 'Doctor visit', 'Treatment', 'Hospital stay', 'Medication', 'Lifestyle Change' and 'Feelings'
 """
 EVENT_TYPE_ANSWER = """
     For example for the bulletpoint 'visiting doctor's' you should return 'Doctors Visit'.
@@ -250,7 +251,8 @@ def life_circumstances_prompt(sex):
     The only output should be the location.
 """
 LOCATION_PROMPT = """
-    Here is the bulletpoint for which you should extract the location:
+    Here is the bulletpoint for which you should extract the location.
+    Explain step by step your conclusions your choice of location: 'Home' or 'Hospital' or 'Doctors' or 'Other'.
 """
 LOCATION_ANSWER = """
     For example for the bulletpoints 'visiting doctor's', you should return 'Doctors'.