Use Onedot to process different file formats quickly and easily. This includes formats such as Excel, delimiter-based formats such as CSV, TSV and TXT with different encodings as well as text-based formats such as XML (e.g. BMEcat) with nested hierarchical structures, JSON and EDI-based data formats (e.g. PRICAT, PRODAT). Corrupt datasets and records are recognized and shown separately for validation.
Is relevant information contained in the text but not available in a structured form as individual attributes? Onedot automatically extracts attributes from product titles, descriptions or key selling points. Different patterns are automatically recognized, such as key value pairs, numbers, areas, volumes, units or list of values. The extraction is based on the latest proprietary algorithms in text analysis and Natural Language Processing (NLP).
The supplier usually uses a different categorisation than you do. Use Onedot to automatically assign products from the supplier to your categories. The machine learning algorithms are trained on the basis of the products in your product database. If a categorization proposal is not correct, you can override the software. The software learns from this feedback and increases the precision and recall of the categorisation.
The vendor generally describes product attributes differently than you do. Therefore a mapping of the attributes is necessary. Based on 12 statistical signals measured in the data and the rich Onedot product knowledge graph, the software makes a proposal on the initial mapping. You can then view the mappings and override and correct them if necessary. The attribute mapping is done per supplier.
Once the attribute mapping is defined, the data is transformed from the supplier to your target structure. Three cases are distinguished: 1) product is newly created, 2) supplier product already exists in the product database and can be used to enrich the existing product and 3) product is to be decommissioned because it is no longer listed in the supplier catalog. Supplier products are matched with your existing products using a unique identifier or a fuzzy matching approach.
Texts, numbers, units, value lists, or booleans should all have a consistent formatting pattern and be consistent across categories and filters. With Onedot, you can apply over 37 predefined data normalization patterns to your product data. These include normalizations such as rounding numbers, defining decimal separators, formatting ranges, mapping text values to list of values and removing HTML markups.
If the same product is provided by different suppliers, we recommend the generation of a Golden Record. In a Golden Record, the product information from the different suppliers is combined in a single data record with maximum information content. Different prioritisation options are available to you to merge records as it’s best for you.
Many products exist in different configuration options. Use sophisticated segmentation algorithms to easily group your products and generate product variants based on the variant-building attributes. This enables you to display the product variants to your customers on the product page. Most frequently used attributes for product variants are color and size, but many other attributes are possible and useful.
After the data preparation process has been completed, the results can be made available in the data format you need. Most of the time, this is a CSV or XML file. The goal is that you can import the results into your product database with a single click.
EASY TO USE
- No query or scripting language to learn.
- Designed with a simple and intuitive interface.
- Built to model human behaviour for complex data processes.
- Constant learning from customer data and user feedback.
- Copes well with changing data structure.
- Developed with sophisticated pattern recognition.
- Highly scalable and future-proof.
- Affordable – pay only for what you need.
- No implementation and maintenance efforts.