NEC Networks & System Integration Corporation - Zoom Meeting Transcription and Scoring Without Code

IchikawaGinji Dataiku DSS Core Designer, Registered Posts: 1 ✭✭

Team members:

  • Ginji Ichikawa
  • Ryotaro Suzuki
  • Takuya Minegishi
  • Junki Toyokawa
  • Yushi Minami

Country: Japan

Organization: NEC Networks & System Integration Corporation

NEC Networks & System Integration provides a wide range of services, from ICT and networking to installation and maintenance, to a variety of customers, including corporations, telecommunications carriers, government agencies, and social infrastructure providers.

Awards Categories:

  • Best Acceleration Use Case
  • Best Moonshot Use Case
  • Best ROI Story

Business Challenge:

Our main business is designing and installing PBX and networks for both private and public customers in Japan. In recent years, we have been also focusing on reselling SaaS in addition to offering system integration and O&M services. One symbolic example would be we are the first Zoom distributor in Japan.

As the first company to bring Zoom to Japan, we needed to continually improve the value of Zoom for our company, so we turned to Dataiku. We needed to attract new customers by offering our existing customers free or low-cost optional Zoom services that other companies did not offer.

Business Solution:

Since we needed to solve the aforementioned issues with members who had limited knowledge of code, Dataiku was a very attractive service that made it easy for even inexperienced programmers to build flows, and easy to implement predictions and analysis. The ability to build, automate, and monitor all in one service made it easy to manage and speedy to develop from verification to deployment.

Only one person actually worked on the construction of the service, using the flow on Dataiku, basically no code, and only referring to external libraries in python as needed. I received advice from several service planners on the finished product, and was able to implement the service in about a month, revising it as needed.

  • Transcription of Zoom recordings

We took Zoom recordings in real-time or by specifying the time, and transcribed them using Open AI's "Whisper" in Dataiku's Python connector.

  • Positive/negative analysis of Zoom transcription function

Positive/negative judgment of meetings was performed by creating a file on Dataiku that divides the zoom transcription results into data sets for each speaker, and passing it through the emotion analysis in the Dataiku plugin.

  • Summary of Zoom transcription results

We have built a function to summarize the results of zoom transcription using Chat-GPT and send it to meeting participants.

  • Dashboard displays results of positive/negative analysis of Zoom's transcription function

The results of positive/negative analysis of Zoom's transcription function are displayed on the dashboard, making it easier to confirm which meetings are negative.

  • Meeting scoring

We acquired Zoom meeting data via API and performed scoring of meetings. We evaluated the meetings on a 10-point scale based on the meeting time and participant information, and tried to derive the optimal meeting, but the results were not very accurate, so we are currently continuing the verification process.

Using Dataiku, we were able to easily form datasets and link them to external libraries using Python connectors. The ability to use Python connectors makes it easy to integrate the output with external services.

Day-to-day Change:

  • By transcribing and summarizing the zoom, it became possible to know the details of the meeting without having to attend the meeting.
  • Positive and negative factors of meetings can be analyzed by analyzing the positives and negatives of meetings.

Business Area: Communication/Strategy/Competitive Intelligence

Use Case Stage: In Progress

Value Generated:

It is envisioned that the process implemented by Dataiku can be implemented to run when the meeting is over, so that there is no need to review the recordings of meetings that they were unable to attend.

Specifically, it is envisioned that managers and above who have two meetings in one day and need to review the recordings will be able to save 2 hours/day by using the transcription to understand the content of the meeting only by using the transcription.

Furthermore, by analyzing the positives and negatives of each meeting for each speaker and graphing them, it was possible to visualize which times of the day there were many positive comments and which times of the day there were many negative comments in the caregiving process.

Our future goal is to reduce meeting time by 1 hour/day per person by scoring meetings to identify unnecessary meetings.

Value Brought by Dataiku:

Dataiku has brought great value to our organization in a variety of ways. In addition to predictive models, the Dataiku platform allows us to implement a wide range of services, from dataset molding to code environments such as Python.

Using Dataiku, we are able to quickly build services and go from planning to release with a great deal of speed. We were also able to broaden the scope of our ideas and propose a number of business ideas using data.

Value Type:

  • Improve customer/employee satisfaction
  • Reduce cost
  • Save time

Value Range: Hundreds of thousands of $

Setup Info
      Help me…