From the perspective of spatial databases

ArticleDirectory

• Recent research on spatial databases
• How to do research

Snow outside, the weather should be the most suitable for reading or writing.

Recent research on spatial databases

In fact, it is not enough to study, but to read articles and books. The system learning of spatial databases started in last December. The reading track is roughly as follows:

(1) Spatial Database (a tour) published in China, this book should be a very good book and I strongly recommend it;

(2) I read some foreign articles, including the earliest R-tree article on ACM, which seems to be the background of CS science;

(3) I downloaded two books from verycd, one of which is mainly a space database, and the other is a semantic web, including a lot of information about the time space database;

Therefore, we probably figured out the system and research context of the spatial database. From the spatial data model to the definition of spatial data and the design of interfaces, the underlying implementation is compared, such as spatial index, query optimization.

I read the postgis manual this evening, but there is still some distance from a product. So I went back and browsed the ArcSDE document on the ArcGIS installation disk. Including the above articles and postgis and ArcSDE notes, and then write them again.

How to do research

To be honest, I have learned how to do research, or how to do research in the CS field. In the past, according to Microsoft Research InstituteResearch on technologies 5-10 years laterIt seems like nothing. In fact, in the major field of CS, it probably includes the basic research part of GIS, which is far from productization.

For example, if you are using a spatial database, you can study the spatial data model at the upper level and the design of the query language. The basis is set theory and mathematical logic, rather than just talking about it. Or the underlyingAlgorithm. The real product is indeed a result of a long time ago, such as spatial index. In postgis, B tree, r tree, or gist can be used, but the document says there are still problems with R-tree; grid is recommended for ArcSDE. Both of them are "primitive" methods.

Therefore, the skill and foundation required for basic research are indeed extraordinary. Cutting-edge fields will be very specialized. For example, for R-tree and its variants, we will study the improvement of an index algorithm under certain circumstances.

Last year I wrote a story about how to splash some research into geographic information science. Now, some of my views are also problematic.

Before doing GIS, I had to do geology for many years, and then did development, and then GIS.For the original professional geology or structural geology, due to regional reasons, in fact, the research is relatively good, for domestic, as long as the tracking of the international frontier, foreign research to move to the domestic is basically enough. The original theories and methods and technologies are basically rare because of the maturity of their disciplines.

Most GIS applications, that isIf GIS is applied to a certain field as a tool and means, the research should be based on other disciplines rather than GIS.. In the software industry, it is the importance of industry knowledge.