Intelligent Supplier Segmentation and Personalized Bidding Recommendation System Based on RFM Score

Abstract

This invention discloses a computer-based bidding recommendation system that collects and stores bid records from vendors on an electronic bidding platform. It calculates an RFM score based on the most recent bid interval, the number of active bidding days, and the number of valid bids, and then groups these bids using a clustering algorithm. The system sets personalized strategies for each group, adding visual icons and filtering functions such as "Previously Bid," "Previously Won a Bid," and "Never Bid" to the case list. It also integrates communication channels such as newsletters or customer service follow-up to improve vendor participation and bidding matching efficiency. R (Recency): defined by the number of days since the most recent bid; F (Frequency): defined by the number of different dates on which bids were submitted within the observation period; M (Monetary): defined by the number of valid bids within the observation period. The scores and groups can be updated periodically to allow the platform to continuously adjust its recommendation and operational strategies.

Claims

1. A smart supplier segmentation and personalized bidding recommendation system based on RFM scores, comprising: a bidding activity database storing bidding activity records of multiple vendors, each bidding activity record including at least one vendor identification code, one inquiry case identification code, one quotation timestamp, and one quotation status field, the quotation status field indicating whether the quotation is a valid quotation or a reply without price; a vendor database storing vendor identification information, contact information, and segmentation tags for each vendor; an RFM scoring engine coupled to the bidding activity database for calculating the Recency (R) score, Frequency (F) score, and Monetary (M) score for each vendor based on the bidding activity records; and a segmentation engine connected to the RFM scoring engine, the segmentation engine being configured to segment the multiple vendors into multiple clusters based on the R, F, and M scores of each vendor, and update the corresponding vendor segmentation tags to the vendor database; A grouping rule repository stores personalized parameters corresponding to multiple grouping tags, including interface presentation parameters and communication strategies; an interface personalization engine is connected to the grouping rule repository and, via the grouping rule repository, to a vendor database, to obtain the vendor's grouping tags when the vendor logs into a bidding platform, and to generate a personalized list of quotations for the vendor based on the corresponding personalized parameters, and to display at least one visual marker for each quotation in the personalized list of quotations; and a communication engine is connected to the grouping rule repository and, via the grouping rule repository, to the vendor database, and is configured to initiate external communication with the vendor based on the communication strategy corresponding to the vendor's grouping tags.

2. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in Request 1, wherein, The RFM scoring engine is configured to map the number of days Δ between a system reference date and the timestamp of the vendor's most recent quote to the Recency score R according to the following intervals: R=7 when Δ ≤ 15 days, R=6 when 15 < Δ ≤ 30 days, R=5 when 30 < Δ ≤ 45 days, R=4 when 45 < Δ ≤ 60 days, R=3 when 60 < Δ ≤ 75 days, R=2 when 75 < Δ ≤ 90 days, and R=1 when Δ > 90 days.

3. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in Request 1, wherein, The RFM scoring engine is configured to: within a pre-defined observation period, use the number of different calendar dates on which each vendor submits at least one quote as the basis for calculating the Frequency score F, and divide the distribution of the number of different calendar dates for all vendors into multiple intervals, assigning the corresponding Frequency score F according to the interval to which each vendor falls; and use the number of valid quotes submitted by each vendor during the observation period as the basis for calculating the Monetary score M, and divide the distribution of the number of valid quotes for all vendors into multiple intervals, assigning the corresponding Monetary score M according to the interval to which each vendor falls.

4. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in Request 1, wherein, The clustering engine is configured to perform an unsupervised clustering algorithm on the feature vector composed of the R, F, and M scores to cluster the multiple vendors into multiple clusters; the clustering rule store maps the clusters to multiple clustering labels, which include at least: important value customers, important retention customers, important development customers, important retention customers, general value customers, general retention customers, general development customers, and general retention customers.

5. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in Request 1, wherein, The interface personalization engine is configured to: for each inquiry in the personalized inquiry list, determine whether the vendor, within a defined backtracking period: (1) submitted any quotation related to the inquiry; (2) won a bid for one of the inquiry related to the inquiry; and (3) submitted a non-price response indicating no price quotation for the inquiry; and display the inquiry as follows: when condition (1) is met, display a first illustration; when condition (2) is met, display a second illustration; and when condition (3) is met, display a third illustration.

6. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in claim 5, wherein, The interface personalization engine is further configured to provide one or more filtering controls, enabling the vendor to narrow the personalized list of requests to requests related to at least one selected illustration.

7. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in Request 1, wherein, The segmentation rule repository defines different communication strategies for value customers, retention customers, development customers, and customer retention customers. The communication engine is configured to select from multiple external communication channels, including at least email and human customer service, based on the communication strategies. For at least one segmentation tag for value or development customers, a recommendation newsletter is periodically sent to the corresponding vendor via email. For at least one segmentation tag for retention customers, customer service follow-up tasks related to the corresponding vendor are generated.

8. The intelligent supplier clustering and personalized bidding recommendation system based on RFM scores as described in claim 1, further comprising a periodic model update scheduler configured to trigger a model update process at a scheduled time interval. This model update process recalculates the R, F, and M scores from the bidding activity database, re-clusters multiple vendors using the clustering engine, and updates the clustering labels stored in the vendor database. In one embodiment, the model update process operates on an analytical database that synchronizes data from the official database and writes the updated RFM scores and clustering labels back to the official database.

9. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in claim 1, wherein, The bidding activity record is generated in response to the interaction between the manufacturer and an electronic bidding platform. The electronic bidding platform allows the manufacturer to submit a valid bid including a unit price for each inquiry, or to submit a non-price response indicating no bid. The RFM scoring engine considers both the valid bid and the non-price response as bids when determining the time of the manufacturer's most recent bid and counting the number of different calendar dates on which the manufacturer submitted at least one bid within an observation period.

10. The intelligent supplier segmentation and personalized bidding recommendation system based on RFM scores as described in claim 1, wherein, The interface personalization engine is configured to apply different preset filtering settings to different segmentation tags. These different preset filtering settings include: for the segmentation tags of value customers, a first preset filtering that emphasizes inquiries similar to previously won bids; and for the segmentation tags of development customers, a second preset filtering that emphasizes inquiries similar to inquiries that have previously submitted bids but have not yet been won.

Images