Please enable JavaScript in your browser.

fltech - 富士通研究所の技術ブログ

富士通研究所の研究員がさまざまなテーマで語る技術ブログ

Introducing Smart Enterprise Data Assistant

Hello, I'm Zhongguang Zheng from our Artificial Intelligence Laboratory. Today, I would like to introduce an exciting new AI core engine, - Smart Enterprise Data Assistant – which is now available from Fujitsu Kozuchi.

This article is part of a series introducing the AI core engines of Fujitsu Kozuchi, and you can find a full summary of the previous articles at the end.

The Smart Enterprise Data Assistant enables on-demand business data analysis and visualization, using chat mode without requiring programming or customization. It is one of Fujitsu Kozuchi’s AI core engines, which enables the fast testing of cutting-edge AI technologies developed by Fujitsu.

Analyzing business data is necessary in a variety of situations. For example, in a retail store, it's necessary to perform sales analysis and analysis of customer behavior. However, advanced analytical and programming skills are required to analyze business data. You need to organize and aggregate data, conduct advanced statistical analysis, and visualize the data, all of which tasks require time. Therefore, it's challenging for every business professional easily to perform data analysis when needed. To overcome this issue, we have developed the Smart Enterprise Data Assistant.

Benefits of Smart Enterprise Data Assistant and how to use it

The principal value of the Smart Enterprise Data Assistant is to perform real-time business data analysis and visualization through chat without the need for programming or customization.

I would like to introduce you to the Smart Enterprise Data Assistant’s capabilities, using the example of sales data analysis in a supermarket.

First, the user inputs the data files. In this example, there are three data files. Commodityinfo.csv includes product data such as product names and selling prices, Customerinfo.csv includes customer data such as gender, age, and store entry time, and CustomerPOSData.csv includes purchased product data. The currently supported data format is CSV.

Using the Smart Enterprise Data Assistant requires two steps. The first step is understanding the content of the data files, and the second step is the actual analysis. I will begin by explaining the first step, involving understanding the content.

The user starts by pressing the analyze files button. The Smart Enterprise Data Assistant analyzes the input files to understand what kind of data they have and what they could potentially be used for, before presenting this information to the user.

In this step, if the Smart Enterprise Data Assistant misunderstands the data, it can impact the next step of data analysis. Therefore, it's crucial for the user to check it thoroughly. If the Smart Enterprise Data Assistant makes a mistake or needs additional information, the user makes a request through chat. Here, I will introduce a situation where additional information is needed.

In the purchased product data file, each product data is listed in one line. Therefore, if the same customer buys multiple products, there will be multiple lines of data. This can be entered in the chat. There are no rules for how to describe it, and you can enter it in natural language. The Smart Enterprise Data Assistant modifies the file description in response to correction requests from users. If there are no issues with the file description provided by the Smart Enterprise Data Assistant, you can proceed to the next step of the analysis after clicking the "The AI has fully understood the uploaded files” check box.  

Let's take the example of analyzing fruit purchasing. If you enter "I want to find customers who bought apples" in the chat, a file of customers that bought apples is outputted .

Next, suppose you want to know the gender of customers who bought fruit including apples. In the entered product data, product names such as apples and bananas are written, but "fruit" is not. The Smart Enterprise Data Assistant matches the concept of fruit with the product names in the entered data file and outputs the results. This allows you to analyze from the perspective of fruit, even if there is no fruit classification in the data file.

Furthermore, the Smart Enterprise Data Assistant has various visualization functions. In this example, you would enter "Please show the gender distribution of customers who bought fruit" and the Smart Enterprise Data Assistant would display a graph like the one below.

Features of Smart Enterprise Data Assistant technology

The Smart Enterprise Data Assistant is a technology that uses Large Language Models (LLM) and generative AI technology to provide real-time data analysis results on demand. It allows for interactive file analysis and fuzzy query analysis, with the following technical features.

・ Understanding and analyzing data through interactive exchanges The Smart Enterprise Data Assistant can understand the csv files in an interactive manner. The Smart Enterprise Data Assistant will first analyze the content of csv files such as header and value information, and then explain the understanding of this information. The csv information can help the Smart Enterprise Data Assistant to find co-relations between different csv files and correctly analyze the user query. The Smart Enterprise Data Assistant can also revise the csv understanding according to the user’s feedback.

・Understanding fuzzy query on basis of the csv data The Smart Enterprise Data Assistant can map the concept in the query to the actual values in the csv data. For example, it can map the concept “fruit” in the query, and “find customers who bought fruit” in retail scenarios to certain fruit names in the csv data, ensuring the analysis accuracy

Interested in testing the Fujitsu Kozuchi?

The Smart Enterprise Data Assistant can help people who want to analyze their own csv data without needing to have data analysis skills. Anyone with data analysis requirements such as business managers and company operators can try the Smart Enterprise Data Assistant . All you have to do is prepare your data in the one header line csv format and upload the files to the Smart Enterprise Data Assistant to start the analysis.

For a demonstration or to test our Smart Enterprise Data Assistant, please contact us here :

In addition to the Smart Enterprise Data Assistant, you can find out more about the other Fujitsu Kozuchi AI core engines on our TECH BLOG.