Subset filter for memory key value table

Abstract

The invention discloses a subset filter for a memory key value table. The subset filter is mainly divided into five parts: definition of the subset filter, data item storage of the subset filter, data item updating of the subset filter, reconstruction of the subset filter and scanning of the subset filter. According to the method, filtering of any type of index data and filtering of any specified data feature are supported, so that a flexible scanning mode is provided, all index data do not need to be traversed during scanning, and the scanning operation is effectively accelerated. The method can be applied to most memory key value tables and has certain universality.

Description

technical field [0001] The invention relates to the financial field requiring high-frequency real-time information processing and storage, such as risk control and anti-fraud, and relates to a subset filter oriented to a memory key-value table. Background technique [0002] In the financial field that requires real-time calculation such as risk control and anti-fraud, it is not only necessary to calculate financial indicators such as "the transaction volume of an entity in the past 1 day" and "the maximum transaction amount of an entity in the past 1 week", but also needs to be calculated such as " Calculation and processing of complex events such as looking for entities that conduct 3C transactions within 30 minutes after recharging the phone bill, and then conduct valuables transactions within 10 minutes”, “find two adjacent transaction entities with a total amount of more than 1 million yuan”, etc. The indicator data of the transaction entity needs to be scanned, and the ...

AI Technical Summary

Problems solved by technology

[0006] In the complex time processing and calculation scenarios of financial risk control and anti-fraud, the scanned event and index data are becoming more and more complex, and the scanning methods are becoming more and more flexible and diverse. However, the above-mentioned indexing method is difficult to meet these scanning requirements. There are the following two Point defect:

[0007] (1) The data types supported by the index are limited: the data format is limited to common data types such as strings and integers, which cannot meet the indexing requirements of complex events and calculation indicators

[0008] (2) The scanning methods based on index support are limited: the prefix index based on dictionary tree only supports prefix scanning of strings, while the index based on B+ tree only supports range scanning of specific ordered data, complex events and calculation indicators that meet specific conditions scanning, which is very inefficient

[0009] To sum up, the existing index-accelerated scanning methods mentioned above cannot fully meet the scanning needs of financial real-time computing and complex event processing such as risk control and anti-fraud.

Images