notesum.ai

Published at December 10

Automating Business Intelligence Requirements with Generative AI and Semantic Search

cs.SE

Released Date: December 10, 2024

Authors: Nimrod Busany¹, Ethan Hadar¹, Hananel Hadad¹, Gil Rosenblum¹, Zofia Maszlanka², Okhaide Akhigbe³, Daniel Amyot³

Aff.: ¹Accenture Labs; ²Avanade; ³EECS, University of Ottawa

Arxiv: http://arxiv.org/pdf/2412.07668v1

Business Question (User): “Please provide the total amount of earnings per product sold in Euro”. Listing 1: Query Generation Response (System) ⬇ SELECT Product.ProductNumber, SUM(SalesOrderDetail.LineTotalCurrencyRate.AverageRate) AS TotalEarnings FROM Product JOIN SalesOrderDetail ON SalesOrderDetail.ProductID = Product.ProductID JOIN SalesOrderHeader ON SalesOrderDetail.SalesOrderID = SalesOrderHeader.SalesOrderID ... WHERE CurrencyRate.ToCurrencyCode = "Euro" GROUP BY Product.ProductNumber
Query Explanation Response (System): The results display the total earnings of each product, calculated by multiplying the sales line total by average currency exchange rate, specifically for transaction converted to Euros, with the data organized by product names.
Execution Results and Test Case (System): The query execution results are displayed in the table in See Figure 2 (left). The table includes the two columns ProductNumber and TotalEarnings, and total earnings per product. Listing 2: Ontology Snippet (System) ⬇ Class specialofferproduct { Data Property: rowguid VARCHAR, Data Property: ModifiedDate VARCHAR, Data Property: SpecialOfferID INT, Data Property: ProductID INT, Object Property: has_product REFERENCES product } Class productdocument { Data Property: ModifiedDate VARCHAR, Data Property: ProductID INT, Data Property: DocumentNode INT, Object Property: has_product REFERENCES product } ... Listing 3: Bindings Snippet (System) ⬇ Class c: specialofferproduct => Table t: specialofferproduct, Table t2: product c.rowguid => t.rowguid, c.SpecialOfferID=> t.SpecialOfferID, c. ProductID =>t. ProductID, c.ModifiedDate=>t.ModifiedDate , c.has_specialoffer=>t.specialofferproduct, c.has_product=> t2.specialofferproduct ...

Business Question (User): “Please provide the total amount of earnings per product sold in Euro”. Listing 1: Query Generation Response (System) ⬇ SELECT Product.ProductNumber, SUM(SalesOrderDetail.LineTotalCurrencyRate.AverageRate) AS TotalEarnings FROM Product JOIN SalesOrderDetail ON SalesOrderDetail.ProductID = Product.ProductID JOIN SalesOrderHeader ON SalesOrderDetail.SalesOrderID = SalesOrderHeader.SalesOrderID ... WHERE CurrencyRate.ToCurrencyCode = "Euro" GROUP BY Product.ProductNumber

Query Explanation Response (System): The results display the total earnings of each product, calculated by multiplying the sales line total by average currency exchange rate, specifically for transaction converted to Euros, with the data organized by product names.

Execution Results and Test Case (System): The query execution results are displayed in the table in See Figure 2 (left). The table includes the two columns ProductNumber and TotalEarnings, and total earnings per product. Listing 2: Ontology Snippet (System) ⬇ Class specialofferproduct { Data Property: rowguid VARCHAR, Data Property: ModifiedDate VARCHAR, Data Property: SpecialOfferID INT, Data Property: ProductID INT, Object Property: has_product REFERENCES product } Class productdocument { Data Property: ModifiedDate VARCHAR, Data Property: ProductID INT, Data Property: DocumentNode INT, Object Property: has_product REFERENCES product } ... Listing 3: Bindings Snippet (System) ⬇ Class c: specialofferproduct => Table t: specialofferproduct, Table t2: product c.rowguid => t.rowguid, c.SpecialOfferID=> t.SpecialOfferID, c. ProductID =>t. ProductID, c.ModifiedDate=>t.ModifiedDate , c.has_specialoffer=>t.specialofferproduct, c.has_product=> t2.specialofferproduct ...