Learning heterogeneous graph embedding for Chinese legal document similarity

By most contemporary standards the document is an object (physical or electronic) on which information is recorded. It would thus have two dimensions, the medium and the content. But this dual presentation is insufficient: it obscures the social function that lends the documentary function to both medium and contents. A good illustration of this ambiguity can be found in the legal framework for information technology of Quebec.4 Quebec law is interesting in this respect because, it tries to define a document beyond the medium it uses by paying attention to information. We can read in Article 3 of the 2001 law this definition: Information inscribed on a medium constitutes a document. The information is delimited and structured, according to the medium used, by tangible or logical features and is intelligible in the form of words, sounds or images. On the face of it, this passage defines a document only in terms of its medium and of its contents. These contents, moreover, are viewed as independent of the medium. But the appearance is deceptive. On the one hand, it is precisely because the document has a function-that of transmission of evidence-that we need a law to define it. We must, indeed, be sure that the object we are talking about will perform this function in the new digital environment. On the other hand, it is indeed because the content can pass from one medium to another that Quebec has tried to define in law the link between one and the other to ensure that the documentary function is preserved.