Gamalon makes information usable

Gamalon Structure: Increasing sales and reducing churn with a clean, structured product catalog

Gamalon has processed data for a variety of e-Commerce platforms and manufacturing companies. A common pain point is that they need to gather data from hundreds of inventory databases, supplier data feeds, web content, and other data sources to see what products are available, in stock, and at what price.

A row from a store’s point-of-sale inventory database might look like this:

ID Description
9140 DIET COKE 2/12/12Z C

You can see that there are two twelve packs of Diet Coke in 12oz cans, and the “C” also helps to indicate a case, i.e. 24 cans. Because data comes from different sources, is expressed in different ways (e.g. milliliters versus ounces), gets re-ordered, abbreviated, misspelled, truncated, deleted, reformatted, etc., the same product might look very different in another database:

ID Description
130597 DT COKE REG 355ML CSE

But really, these are both the same product, and we wish that both product descriptions were correctly structured into a format like this, with usable columns:

Product Volume Unit Container Quantity
Diet Coke 12 Ounce Can 24

In the past, it has taken a time-consuming and expensive combination of people and manually driven software to process unstructured data – there has never really been a great solution. Now, Gamalon Structure autonomously structures text data using Gamalon’s Bayesian Program Synthesis (BPS) technology.

Here is another example of unstructured data that Gamalon Structure can successfully process:

Dell XPS 17.3 L701X Microsoft Windows Home Premium 64X. Intel core I5M460 @ 2.53 GHz, 2534 Mhz, 2 cores 4 Processors, 8GB Ram, 7.4GB Physical memory, Nvidia Geforce GT435 Graphics Card. Amazing Sound system. 500GB Hard drive, Has Extra Hard Drive Slot. Needs New Battery and Screen, Screen has some Lines on the Left side, only somewhat annoying.

Which is structured to:

Product Size Model Processor Clock Speed RAM Hard Drive Graphics Card
Dell XPS 17.3 in L701X i5 M460 2.53 GHz 8 GB 500 GB NVIDIA GT435M

Gamalon Match: Doubling average basket size for e-Commerce

Once data has been structured, it can be matched, linked, and de-duplicated.

In this example video, Gamalon Match takes ten Coca-Cola products (left column “Master File”) and matches them to a store database (middle column “Vendor File”). Gamalon does a great job of matching the correct products, even when their descriptions are highly messy and abbreviated. The confidence of the Gamalon system evolves as it examines the data (right column “Match Probabilities”).

Increasing the number of products that are available in an e-Commerce platform, especially “long-tail” high-margin products, can nearly double the average order!