Using the app

Upon installing the native app, a new database is created in Snowflake under the application name you chose during installation. This database includes a schema named datadotworld_api, containing a range of functions and stored procedures that simplify the use of data.world functionality.

When you open a worksheet in Snowsight or launch a Streamlit-in-Snowflake app, you can expand this database to view the functions and their signatures.

The functions and procedures from the native app can be easily called from a Python Worksheet or Streamlit-in-Snowflake app inside Snowsight. 

Following is an example call to the ask_question function in a Streamlit:

with st.form("chat_tab_form"):
   st.info("Enter a question to interact with data.world's AI context engine.", icon="ℹ️")
# give the user a radio option to toggle demo mode
 demo_mode = st.radio("Demo Mode", ["On", "Off"],
 key="chat_tab_demo_mode", index=1, help="Toggle demo mode to see the AI context engine in action")
    question = st.text_area("Question", key="chat_tab_question",
    help="Enter a question to ask the AI context engine")
    submitted = st.form_submit_button("Submit Question")
 if submitted:
 # submit the question
dataset_id = app_state["selected_dataset"]
org_id = app_state["selected_org"]
demo_mode = True if demo_mode == "On" else False
query_sql = f"""SELECT datadotworld_api.ask_question(
owner => ? ,            			
project_id => ? ,             
nl_question => ? ,            			
demo_mode => ?        			
)"""       
response = session.sql(query_sql, params=[owner,
project_id, nl_question, demo_mode]).collect()[0][0]

# convert the response to a dict 
answer_response = json.loads(response)

# display the results to the userx                    
answer = answer_response.get('answer')                    
st.subheader('AI Context Engine Response')                    
st.write(f"**Answer:**")                    
if 'error' in answer.lower():                        
.error(f'Error: {answer_response.get("answer")}')                        
return                    
st.caption(answer)                    
st.write('**SPARQL Query:**')                    
st.code(answer_response.get('sparql'), language='sparql')                    
if answer_response.get('terms'):                        
st.write('**Terms Used:**')                        
terms = pd.DataFrame(answer_response.get('terms'))                        
terms = terms.rename(columns={'term': 'Term', 'definition': 'Definition', 'iri': 'data.world Link'})                        
# change column order                        
terms = terms[['Term', 'Definition', 'data.world Link']]                        
st.dataframe(terms)

Functions can also be from SQL worksheets inside Snowsight. 

Following is an example call to the ask_question API from the worksheet:

-- ask question
SELECT PARSE_JSON(DATADOTWORLD_PLUGIN_APP.DATADOTWORLD_API.ASK_QUESTION(
    owner => 'global',    
    project_id => 'chat-with-our-data',
    nl_question => 'How many users do we currently have?',
    demo_mode => True
));

Along with the Functions, the application contains a Streamlit based UI that can be accessed from Snowsight.

To access the app: