tweak some prompts

Author: sunu

geodini/agents/simple_geocoder_agent.py

@@ -53,9 +53,14 @@ class RerankingResult:
         
         From the results list, return a JSON object with:
         1. "most_probable": The ID of the most relevant result
-        2. "next_probable": A list of IDs of the next 3 most relevant results in order of relevance
+        2. "next_probable": A list of IDs of the next 5 most relevant results in order of relevance
 
         Make sure the returned IDs are in the results list.
+
+        While reranking, consider the following:
+        - The query might be a shortened name and the result might be a full name. For example, "United States" or "United States of America" is a match for "USA" or "The US".
+        - The query might be a informal name and the result might be a formal name. For example, "District of Columbia" is a candidate for "DC" or "Washington, D.C." or even just "Washington". So consider all possible variations of the query when matching.
+        - Consider geographical context. For example, "London in Canada" should rank "London, Ontario" higher than "London, England" because it is more likely to be the correct answer.   
 """,
 )
 
@@ -184,13 +189,13 @@ async def search_places(query: str) -> list[Place]:
             for place_id in reranked_results.output.next_probable
         ]
     else:
-        most_probable = []
+        most_probable = None
         next_probable = []
 
     return {
         "most_probable": most_probable,
         "next_probable": next_probable,
-        "results": list(results_dict.values()),
+        "results": [most_probable, *next_probable],
         "query": query,
         "rephrased_query": rephrased_query.output.query,
         "country_code": rephrased_query.output.country_code,

geodini/agents/utils/geocoder.py

@@ -47,9 +47,6 @@ def geocode(query: str) -> list[dict[str, Any]]:
 
     try:
         with engine.begin() as conn:
-            if limit is not None:
-                result = conn.execute(text(sql_query), {"query": query, "limit": limit})
-            else:
             result = conn.execute(text(sql_query), {"query": query})
 
             rows = result.fetchall()
@@ -125,8 +122,8 @@ def build_postgis_query() -> str:
                 WHEN 'macroregion' THEN 2.0
                 WHEN 'region' THEN 2.0
                 WHEN 'macrocounty' THEN 2.0
-                WHEN 'county' THEN 1.1
-                WHEN 'localadmin' THEN 1.5
+                WHEN 'county' THEN 1.0
+                WHEN 'localadmin' THEN 1.1
                 WHEN 'locality' THEN 0.9
                 WHEN 'borough' THEN 0.8
                 WHEN 'macrohood' THEN 0.8
@@ -164,3 +161,10 @@ def build_postgis_query() -> str:
     if postgis_results:
         print("\nSample PostgreSQL result:")
         pprint(postgis_results[0])
+
+    # print the name and similarity score of all results
+    for result in postgis_results:
+        print(
+            f"Name: {result['name']}, Similarity: {result['similarity']}, Type: {result['subtype']}, Country: {result['country']}"
+        )
+    print(len(postgis_results))