Data transmission and processing shift to a higher gear
The QuiXProc technology, developed by Innovimax in partnership with the MOSTRARE research team, can be used to quickly compile and return data exchanged over the Web under the XML standard.
Data and document processing as part of exchanges between servers is a major challenge for companies. Indeed, their activities now depend on exchanges of data and any slowing down in processing can have direct repercussions on the progress of their projects.
These slowdowns are due to the multitude of operations required as part of the exchanging and processing of the data. In the simplest cases, the stages are as follows: check that the data complies with the expected schema (XML Schema), then extract part of the data (XPath and/or XQuery) and process it into the target format (XSLT). Obviously, the sequences of operations handled by QuiXProc are usually far more complex. It is then necessary to apply an organisation process known as ‘orchestration’. The orchestration is described by the XProc language recommended by the W3C.
Another major constraint involves processing time and the memory capacity of machines. Processing can be performed using an XML tree structure but this requires loading the data to memory which implies that the memory must be big enough to contain the tree structure. Imagine, for instance, how much memory the millions of parameters generated by the World Bank might take up!
The Mostrare research team and Innovimax have been working together, focusing on the XProc language, to propose an efficient implementation solution based on the assumption that all the data that needs to be processed, whatever the context, is being transmitted in a continuous stream (also known as streaming). A stream is the translation of an XML tree into a text file. It consists of data and opening and closing tags which describe the structure of the XML tree. QuiXProc orchestrates this process and performs all the operations directly on the stream, instead of handling the data in an XML tree. This solution can speed up the exchange and processing of data while reducing the amount of data that must be loaded into memory.
The promising results of this technology transfer have encouraged the Mostrare research team and Innovimax to continue their work in order to fine-tune the efficiency of the processing and exchanging of data.
These articles could interest you:
A technology created from the partnership between Inria and Innovimax.
Innovimax, which was formed in 2006, advises and assists companies in industrialising their documentary processes. Innovimax, which since 2008has been working in close cooperation with several entities of Inria as part of an ANR project, has formed a close cooperation with the MOSTRARE project team of Inria Lille by implementing a bilateral technology transfer partnership based on the QuiXProc project.
As part of this partnership, Innovimax has implemented a CIFRE thesis and has contributed its know-how as well as its close involvement with the W3C and ISO standards, while Inria has contributed research and legal resources. The Institute’s input also enabled the recruitment of a transfer and innovation engineer to ensure the proper running of the project. The quality of the partnership is already yielding significant results.
The XML language
XML (Extensible Markup Language ) is a language in which data is formatted between tags that are used to separate off any reference information (family name, first names, address, product reference numbers, etc.) or packets of information that a web page can contain. The result is usually presented as a tree structure of tags.