Today, most ERP and e-commerce system data is stored as structured data because they are built with relational database technology (RDB). On the other hand, especially those applications and systems developed through mobile application (Mobile App) technology in recent years, the application of unstructured data structure is becoming more common.
Over the years, we have developed and reviewed many ERP and e-commerce systems. As far as we know, many systems designed and developed a few years ago also contain a lot of "unstructured data" – although the system is basically developed by RDB, the data collected and stored in the system (in RDB terms: Tables, Columns and Rows of data) is not correlated, or we can say that the data does not have a predefined data model. Many CEOs, CFOs or senior managers now find that it is not easy to discover business intelligence in huge data records, and various departments may have to pay unrealistic costs to achieve such a goal.
It is true that many companies already have systems that store large amounts of unstructured or structured data, but these data have not been properly capitalized on their potential.
Recently, we completed an enhancement project of an ERP and e-commerce system with an online survey application. The survey data collected by the system is "unstructured"– just a lot of text, numbers, choice preferences etc. You can imagine it as a single spreadsheet: each column being filled with text and numeric answers, and each row containing a set data per customer. A few years ago, it was difficult to infer any valuable business intelligence from hundreds or thousands of rows and columns (i.e. customer survey answers) inside such a system, until the advent of using machine learning (ML) technology.
As an AI developer in this project, we wrote an embedded module in the ERP and e-commerce system to collect, analyze and report "unstructured" data. Our AI module and support service cover the following tasks:
* Clean and standardize data
* Continuous data training and optimization of ML hyper-parameters
* Use machine learning model to classify and predict data
* Regular management reporting
After applying ML technology to the existing ERP and e-commerce system, the customer management team can obtain more valuable information and reports faster, and make accuracy assessment easier.
If you want to know more about how to improve your business, please contact us for further discussion.
Related reading Wiki – Unstructure data