In this Quickstart guide, you will be help the fictitious food truck company, Tasty Bytes, to identify where their customer experience may be falling short at the truck and business level by leveraging Snowflake Cortex within Snowflake Notebook. They collect customer reviews to get customer feedback on their food-trucks which come in from multiple sources and span multiple languages. This enables them to better understand the areas which require improvement and drive up customer loyalty along with satisfaction.


What You'll Need

You will need the following things before beginning:

What You'll Learn

In this quickstart, you will learn:

What You'll Build


You will use Snowsight, the Snowflake web interface, to:

Creating Objects, Loading Data, and Joining Data


You will use Snowsight, the Snowflake web interface, to create Snowflake notebook by importing notebook.

Enabled by Cortex Translate

You will leverage Translate - one of the Snowflake Cortex specialised LLM functions are available in Snowpark ML, to translate the multilingual reviews to english to enable easier analysis. This is done within the notebook using following code snippet in cell CORTEX_TRANSLATE.

# Conditionally translate reviews that are not english using Cortex Translate
reviews_df = reviews_df.withColumn('TRANSLATED_REVIEW',when(F.col('LANGUAGE') != F.lit("en"), \
                                                            cortex.Translate(F.col('REVIEW'), \
                                                                            F.col('LANGUAGE'), \
                                                                            "en")) \

reviews_df.filter(F.col('LANGUAGE') != F.lit("en")).select(["REVIEW","LANGUAGE","TRANSLATED_REVIEW"]).show(3)


In this section, you will make use of Snowflake Cortex - Complete to categorise reviews to understand:

Get ratings based on review

Get intention to recommend based on review

Enabled by Cortex Sentiment

So far you saw Snowflake Cortex - Translate & Complete. Next, you will look at another task specific LLM function in Cortex - Sentiment. This sentiment function is used to understand the customer's tone based on the review they provided. Sentiment return value between -1 and 1 such that -1 is the most negative while 1 is the most positive. This is done within the notebook using the following code snippet in cell CORTEX_SENTIMENT.

# Understand the sentiment of customer review using Cortex Sentiment
reviews_df = reviews_df.withColumn('SENTIMENT', cortex.Sentiment(F.col('REVIEW')))["REVIEW","SENTIMENT"]).show(3)

Enabled by Cortex Complete

Taking this analysis a step further, you will be looking at aspect based sentiment instead of just the overall sentiment of review and understand what the customers think about different aspects like food quality, service, pricing etc. This done by leveraging Snowflake cortex - Complete coupled with a prompt that includes one shot example. This is done within the notebook using the following code snippet in cell ASPECT_BASED_SENTIMENT.

# Prompt to understand sentiment for different categories mentioned in the customer review
# We employ one shot incontext learning to inform LLM
prompt = """[INST]
You are analyzing food-truck customer reviews to undertsand what a given review says about different relevant categories like \
food quality, menu options, staff, overall experience, price, ambience, customer support, \
hygiene standards etc and if sentiment is negative,positive or neutral for that category. \
Only answer in a single valid JSON containing "category", "sentiment" and "details". \
Make sure there is no additional text and not mention categories in answer which are not \
talked in the review. \
Get category based sentiment for the follwoing customer review:
"This food truck offers a disappointing experience. \
The menu lacks healthy options and the food quality is subpar. Finding a parking spot near the \
truck can also be a frustrating ordeal. Additionally, the value for money is not worth it. To top \
it all off, the service provided at this food truck is less than pleasant, adding to the overall \
negative dining experience. Tried reaching out the customer support but was unable to get through." 
Answer : [{     "category": "food quality",     "sentiment": "negative",    "details": "subpar quality"   }, {     "category": "menu options",     "sentiment": "negative",     "details": "lacks healthy options"   },   {     "category": "staff",     "sentiment": "negative",     "details": "unpleasant"   },   {     "category": "price",     "sentiment": "negative",     "details": "not worth the money"   },   {     "category": "experience",     "sentiment": "negative",     "details": "regrettable dining experience"   },   {     "category": "customer support",     "sentiment": "negative",     "details": "unable to get through"   } ].
Get category based sentiment for the follwoing customer review:

# Ask Cortex Complete and create a new column
review_df = reviews_df.withColumn('CATEGORY_SENTIMENT', cortex.Complete('mixtral-8x7b', \
                                                            F.concat(F.lit(prompt), \
                                                                    F.col('REVIEW'), \


In this section, you will leverage Snowflake Cortex - Complete to identify the issues that are mentioned in customer reviews and understand:

Issues at business level

Issues at truck level

Take action assisted by Cortex Complete

In the final step, you will utilize Snowflake Cortex - Complete to draft an email to the owner of the most negatively reviewed truck summarizing the issues that are highlighted in customer reviews along with any recommendation to remedy those issues.This is done within the notebook using following code snippet in cell GENERATE_EMAIL_RESPONSE.

# Prompt to get an email draft which reports the main issues with Truck 5 with recommendations to solve
prompt =""" [INST]### Write me survey report email to the franchise owner summarizing the issues mentioned in following \
aggregated customer review with three concise bullet points under 50 words each such that each bullet \
point also has a heading along with recommendations to remedy those issues.###"""+ reviews_agg_str +""" \
Mention the truck brand name and location in the email.[/INST]"""

# Print the result from Cortex Complete
print(cortex.Complete('mixtral-8x7b', prompt))

Congratulations! You've successfully enabled customer review analytics by leveraging Snowflake Cortex within Snowflake Notebook. And all this without ever needing to move any data outside of secure walls of Snowflake or managing infrastructure.

What we've covered

With the completion of this quickstart, you have now:

Related Resources

Want to learn more about the tools and technologies used in this quickstart? Check out the following resources: