Adaptively enhancing procurement data

1. A computer-implemented method comprising:

obtaining, via upload to a computer system or from live data sources, a plurality of unstructured purchase data having as mandatory fields only line item title, date, and spend amount, the plurality of unstructured purchase data being related to purchases of a plurality of items in a plurality of commodity groups and offered by a plurality of suppliers;

for each purchase line item represented in the plurality of unstructured purchase data, the computer system executing: determining a plurality of field-level data quality scores for a plurality of data fields in each purchase line item; weighting the plurality of field-level data quality scores; determining a row-level quality score based on a sum of the weighted field-level data quality scores; based on the row-level quality score, enriching the each purchase line item in the plurality of unstructured purchase data based on product details and attributes obtained from an item master database, to form an enriched purchase dataset;

processing the enriched purchase dataset using a hierarchical classifier using additional semantic data elements to output a series of natural spend clusters corresponding to product categories represented in the enriched purchase dataset;

displaying on a graphical user interface of a computer display device each of the natural spend clusters in a treemap visualization in which each of the natural spend clusters is a first rectangle corresponding to a product category and a plurality of second rectangles corresponding to subcategories of the product category, each of the second rectangles having a size corresponding to an aggregated spend amount of individual purchases of a corresponding subcategory or a number of items purchased;

in response to a first user input via a control device to select a particular cluster in the treemap visualization and to drag the particular cluster to another cluster, combining the particular cluster and another cluster, and automatically morphing the treemap visualization to reflect the combining.

2. The computer-implemented method of claim 1 , further comprising, in response to a second user input via the control device to select the particular cluster in the treemap visualization, causing displaying, in the treemap visualization, a graphical menu of actions that are programmed to rename, delete, edit or assign a taxonomy category to the particular cluster.

3. The computer-implemented method of claim 2 , further comprising recording, in association with the hierarchical classifier, the first user input and the second user input as organization-specific training data for the hierarchical classifier.

4. The computer-implemented method of claim 1 , the processing using the hierarchical classifier further comprising:

processing the enriched purchase dataset using natural language analysis to extract brand, product type, and attributes to form semi-structured product line items;

providing the semi-structured product line items and a label exclusion dictionary to a hierarchical clustering engine, the hierarchical clustering engine being programmed to prepare an index of each purchase line item in the plurality of unstructured purchase data and to identify clusters of purchases represented in the enriched purchase dataset and category labels for the clusters of purchases.

5. The computer-implemented method of claim 1 , further comprising processing the enriched purchase dataset using the hierarchical classifier using the additional semantic data elements to output the series of the natural spend clusters that represent aggregate spending amount of individual purchases in the product categories represented in the enriched purchase dataset, an organization spend taxonomy not being available.

6. The computer-implemented method of claim 1 , further comprising processing the enriched purchase dataset using the hierarchical classifier using the additional semantic data elements to output the series of the natural spend clusters that represent aggregate numbers of items purchased of individual purchases in the product categories represented in the enriched purchase dataset, an organization spend taxonomy not being available.

7. The computer-implemented method of claim 1 , further comprising causing displaying the treemap visualization using a different color in a treemap of the treemap visualization for each commodity type represented in the treemap visualization.

8. One or more computer-readable non-transitory storage media storing one or more sequences of stored program instructions which, when executed using a computer system, cause the computer system to:

obtain, via upload to the computer system or from live data sources, a plurality of unstructured purchase data having as mandatory fields only line item title, date, and spend amount, the plurality of unstructured purchase data being related to purchases of a plurality of items in a plurality of commodity groups and offered by a plurality of suppliers;

for each purchase line item represented in the plurality of unstructured purchase data, the computer system executing: determine a plurality of field-level data quality scores for a plurality of data fields in each purchase line item; weight the plurality of field-level data quality scores; determine a row-level quality score based on a sum of the weighted field-level data quality scores; based on the row-level quality score, enrich the each purchase line item in the plurality of unstructured purchase data based on product attributes obtained from an item master database, to form an enriched purchase dataset;

process the enriched purchase dataset using a hierarchical classifier using additional semantic data elements to output a series of natural spend clusters corresponding to product categories represented in the enriched purchase dataset;

displaying on a graphical user interface of a computer display device each of the natural spend clusters in a treemap visualization in which each of the natural spend clusters is a first rectangle corresponding to a product category and a plurality of second rectangles corresponding to subcategories of the product category, each of the second rectangles having a size corresponding to an aggregated spend amount of individual purchases of a corresponding subcategory or a number of items purchased;

in response to a first user input via a control device to select a particular cluster in the treemap visualization and to drag the particular cluster to another cluster, combine the particular cluster and another cluster, and automatically morph the treemap visualization to reflect the combining.

9. The one or more computer-readable non-transitory storage media of claim 8 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to execute, in response to a second user input via the control device to select the particular cluster in the treemap visualization, cause displaying, in the treemap visualization, a graphical menu of actions that are programmed to rename, delete, edit or assign a taxonomy category to the particular cluster.

10. The one or more computer-readable non-transitory storage media of claim 9 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to record, in association with the hierarchical classifier, the first user input and the second user input as organization-specific training data for the hierarchical classifier.

11. The one or more computer-readable non-transitory storage media of claim 8 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to:

process the enriched purchase dataset using natural language analysis to extract brand, product type, and attributes to form semi-structured product line items;

provide the semi-structured product line items and a label exclusion dictionary to a hierarchical clustering engine, the hierarchical clustering engine being programmed to prepare an index of each purchase line item in the plurality of unstructured purchase data and to identify clusters of purchases represented in the enriched purchase dataset and category labels for the clusters of purchases.

12. The one or more computer-readable non-transitory storage media of claim 8 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to process the enriched purchase dataset using the hierarchical classifier using the additional semantic data elements to output the series of the natural spend clusters that represent aggregate spending amount of individual purchases in the product categories represented in the enriched purchase dataset, an organization spend taxonomy not being available.

13. The one or more computer-readable non-transitory storage media of claim 8 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to process the enriched purchase dataset using the hierarchical classifier using the additional semantic data elements to output the series of the natural spend clusters that represent aggregate numbers of items purchased of individual purchases in the product categories represented in the enriched purchase dataset, an organization spend taxonomy not being available.

14. The one or more computer-readable non-transitory storage media of claim 8 , further comprising sequences of stored program instructions which, when executed using the computer system, cause the computer system to cause displaying the treemap visualization using a different color in a treemap of the treemap visualization for each commodity type represented in the treemap visualization.