Automated Product Data Classification, UNSPSC Classification

Super-fast, Highly Accurate, Automatic Product Data Classification

Zoomix Accelerator is the world's most advanced software for automatically classifying large volumes of complex, semi-structured and "dirty" product data, "out-of-the-box." With absolutely NO manual rule or script development, Zoomix Accelerator can classify challenging product records into one or more taxonomies, including both standard taxonomies (such as UNSPSC, eCl@ss, ETIM and GS1) and custom hierarchies used internally by companies.

The Challenge of Classifying Business Data

• Indeterminate Business Rules – In most modern corporate scenarios, data quantities are vast, variability is extreme, and inconsistencies are commonplace. Given this reality, it becomes virtually impossible to manually develop sets of if-then-else rules to accurately define how to properly classify data records.
• Unstructured Database Information – Much of the data in corporate databases is unstructured or poorly-structured. Important descriptive characteristics of an entity are frequently lumped together in a single descriptive text field, making it extremely difficult for a computer program to determine the meaning of those descriptors. Descriptors are often presented in a variety of different formats, making it that much harder to determine into which category a particular item should be placed.
• Multiple Taxonomies and Languages – Data classification is exponentially more difficult when requirements include classification of data into different taxonomies (perhaps one for internal use and UNSPSC for trading with business partners) or when data sets contain multiple languages (as is becoming more and more common in the global marketplace).
• Multiple Categories within a Hierarchical Taxonomy – Another difficulty in accurate classification arises when the same item should ideally be included in multiple categories within a hierarchical taxonomy. Manual classification efforts may well place an item in one category while overlooking other equally-relevant categories. Automated systems, too, will not necessarily identify more than one relevant category for a particular item.

The Power Of Self-Learning Technology

Self-Learning Technology introduces an entirely different approach to rapidly and accurately classify even the most complex and variable business data. In fact, the "machine learning" approach is the only approach available that delivers the accuracy, speed, and flexibility required for successful classification of real-world business data.

Other approaches to classification rely on extensive "dictionaries" and "rule bases" which attempt to identify the relationships between data entities and the categories in a particular classification schema. However, creating and maintaining these dictionaries to the degree required for effective classification is a time-consuming, expensive, inflexible and almost impossible task.

On the other hand, the approach taken by Zoomix with its Self-Learning Technology provides the enterprise with a number of powerful capabilities, each of which delivers tangible business advantages:

• Automatic Discovery of Business Rules – Very rapid system implementation, highly-accurate results, no time-consuming and expensive dictionary/rules definition phase, automatic ongoing revision/improvement of classification accuracy, new data handled rapidly, without manual development
• Classification of Unstructured Data – Accurately classify all enterprise data regardless of structure/format, avoid expensive and time-consuming data cleansing and formatting undertakings
• Support for Multiple Classification Schemas – Support all company-specific and standard classifications "out-of-the-box," perform classification both for internal and external data (e.g. product data from trading partners), adopt new taxonomies now or in the future without losing classification capabilities or discarding existing schemas
• Support for Customized Classification Schemas – Great flexibility to adapt standard classification schemas to particular business needs, use classification schemas requiring multiple parents for particular categories as well as those which prohibit this characteristic, allow particular categories in the hierarchy to serve as "category headings" which are not intended to actually contain data items
• Classification into Multiple Categories – Increased sales when customers can find products when browsing/searching in any category relevant for the desired product, data items included in all relevant reports and analyses
• Support for Any Type of Data in Any Language(s) – No need to define rules dictionaries for each type of data, no need to define rules/dictionaries for different languages, handles multi-lingual data with ease

Zoomix Accelerator is a Powerful PIM Accelerator Tool

Zoomix Accelerator offers enterprises a fast and highly-automated track to achieving Product Information Management (PIM) data excellence and its resulting business benefits. The product combines advanced semantic and linguistic analysis with machine learning to automatically and accurately classify, match and standardize complex and highly-variable product data. Patent-pending Self-Learning Technology allows the software to learn, understand, and correct any type of complex product data on the fly.

The technology, easily integrated into existing applications and workflows, begins delivering business benefits immediately while obviating the need for large-scale pre-deployment projects. No manual rules development is required, and the software continues to learn and improve its accuracy in the background, by "observing" decisions made by business users as they go about their routine work.

Zoomix Accelerator dramatically enhances any PIM system by providing advanced data correction functionality lacking from these systems, such as:

• Auto-classification to any standard taxonomy (UNSPSC, e-Class, GS1, ETIM or custom)
• Matching and de-duplication within and between different applications and systems, even with wildly varying formats and terminologies
• Accurate attribute extraction from free-text product descriptions
• Consistent and standardized product records from structured and semi-structured data
• Automatic identification of functional equivalents