Engineering Rock Mass Classifications-三种先进的搜索引擎比较

1 引言

这个笔记以搜索"Engineering Rock Mass Classifications"为例，比较了目前我们使用的三种最先进的(SOTA, State-Of-The-Art)搜索引擎得出的结果：(1) Google Scholar(Google 学术搜索); (2) Semantic Scholar(语义学术搜索); (3) GeotechSet+SSGeotech(岩土工程数据集)。 

2 Top 10结果比较

2.1 Google Scholar

Google 学术搜索得出的Top 10结果如下：

(1) Engineering rock mass classifications: a complete manual for engineers and geologists in mining, civil, and petroleum engineering [book]

(2) Applying rock mass classifications to carbonate rocks for engineering purposes with a new approach using the rock engineering system

(3) Tunnel design by rock mass classifications

(4) Engineering rock mass classification [book]

(5) TBM performance estimation using rock mass classifications

(6) Stability assessment of volcanic lava tubes in the Galápagos using engineering rock mass classifications and an empirical approach

(7) rock mass properties–An example comparing Discrete Fracture Network (DFN) modeling and empirical relations based on engineering rock mass classifications

(8) Use of rock mass classifications for design: recommendations and suggestions

(9) Stability assessment of natural caves using empirical approaches and rock mass classifications

(10) Rock mass classification: a practical approach in civil engineering  [book]

2.2 Semantic Scholar

Semantic 学术搜索得出的Top 10结果如下：

(1) Engineering Rock Mass Classifications: A Complete Manual for Engineers and Geologists in Mining, Civil, and Petroleum Engineering

(2) PANEL WIDTH AFFECTED BY ROCK MASS CLASSIFICATIONS (ABU-TARTUR PHOSPHATE MINES)

(3) Most Used Rock Mass Classifications for Underground Opening

(4) ESTIMATING HOEK-BROWN ROCK MASS STRENGTH PARAMETERS FROM ROCK MASS CLASSIFICATIONS

(5) Application of Fuzzy Set Theory to Rock Engineering Classification Systems: An Illustration of the Rock Mass Excavability Index

(6) Examining Feasibility of Developing a Rock Mass Classification for Hard Rock TBM Application Using Non-linear Regression, Regression Tree and Generic Programming

(7) Application of Fuzzy Synthesis Assessment Method on Classification of Rock Mass in Mine

(8) Prediction of the Rock Mass Diggability Index by Using Fuzzy Clustering-Based, ANN and Multiple Regression Methods

(9) Philosophy of Engineering Classifications

(10) Research on rationality of classification of rock masses for the building slope engineering and its application

2.3 GeotechSet+SSGeotech

与上面两个搜索引擎不一样，GeotechSet+SSGeotech是专用的岩土工程数据集，首先从这个两个数据集中把所有的与"Engineering Rock Mass Classifications"相关的文献提取出来，然后使用AI算法进行排序，同时使用了5个预训练模型：

paraphrase-mpnet-base-v2paraphrase-MiniLM-L6-v2nq-distilbert-base-v1allenai-spectermsmarco-distilbert-base-v4

因而得出的结果比Google Scholar和SS得出的结果更加准确，其中包括了一些这二者在Top 10中没有得到的重要结果:

[1] Bieniawski Z T (1976) Rock mass classifications in rock engineering.

[2] Bieniawski, Z.T. (1993) Classification of rock masses for engineering: The RMR system

[3] Rock mass classifications and their engineering utilization

[4] Engineering Application Of Rock Mass Classifications Systems

[5] Comparison of Slope Mass Ratings Classification Systems: A Review

[6] Influence Of Geological And Geotechnical Characteristics During The Redaction Of Tunnel Projects

3 结束语

本质上，上述三种搜索引擎都使用了先进的AI算法，不过由于各自数据集的不同以及使用的具体算法不同，导致了结果出现差异。从专业的角度来看，Google 学术搜索比Semantic 学术搜索得出的结果相对好些，但我们自己的数据集能产生出比这二者更好的结果。SSGeotech目前共有岩土工程论文69,810篇。