Cardinality Estimator:Processing SQL with a Vertical Scanning Convolutional Neural Network

Although the popular database systems perform well on query optimization,they still face poor query execution plans when the join operations across multiple tables are complex.Bad execution planning usually results in bad cardinality estimations.The cardinality estimation models in traditional databases cannot provide high-quality estimation,because they are not capable of capturing the correlation between multiple tables in an effective fashion.Recently,the state-of-the-art learning-based cardinality estimation is estimated to work better than the traditional empirical methods.Basically,they used deep neural networks to compute the relationships and correlations of tables.In this paper,we propose a vertical scanning convolutional neural network(abbreviated as VSCNN)to capture the relationships between words in the word vector in order to generate a feature map.The proposed learning-based cardinality estimator converts Structured Query Language(SQL)queries from a sentence to a word vector and we encode table names in the one-hot encoding method and the samples into bitmaps,separately,and then merge them to obtain enough semantic information from data samples.In particular,the feature map obtained by VSCNN contains semantic information including tables,joins,and predicates about SQL queries.Importantly,in order to improve the accuracy of cardinality estimation,we propose the negative sampling method for training the word vector by gradient descent from the base table and compress it into a bitmap.Extensive experiments are conducted and the results show that the estimation quality of q-error of the proposed vertical scanning convolutional neural network based model is reduced by at least 14.6%when compared with the estimators in traditional databases.

ETH-GQS: An estimation of geoid-to-quasigeoid separation over Ethiopia

The determination of accurate orthometric or normal heights remains one of the main challenges for the geodetic community in Ethiopia.These heights are required for geodetic and geodynamic scientific research as well as for extensive engineering applications.The main objective of this study is to estimate the geoid-to-quasi geoid separation(GQS)in Ethiopia(ETH-GQS).Such separation would be required for the conversion between geoid and quasigeoid models,which is mandatory for the determination of accurate geodetic heights in mountain regions.The airborne free-air gravity anomalies and the topo-graphic information retrieved from the SRTM3(Shuttle Radar Topography Mission of a spatial resolution 3 arc-second)digital elevation model were used to compute the ETH-GQS model according to the Sjoberg's strict formula for the geoid-to-quasigeoid separation.The ETH-GQS was then validated using GNSS-levelling data as well as geoid heights determined from different Global Geopotential Models(GGMs),namely the EGM2008,EIGEN-6C4 and GECO.The results reveal that the standard deviation of differences between the geoid heights obtained from the EIGEN-6C4 model and the geometric geoid heights obtained from GNSS-levelling data were improved by~75%(i.e.from~24 to~6 cm)when considering GQS values obtained from the ETH-GQS.