Ceci est une ancienne révision du document !
Table des matières
Advanced use of Gephi
Gephi is a platform exploration that allows interactive data visualization for analysis of all types of networks. It is an (free) open-source software, written in Java and which operates on Windows, Linux and Mac OS X. It uses a 3D rendering engine to obtain a flexible architecture observable data in space. Patent2net uses Gephi in order to visualize patents space and to see how they are linked through specific networks.
Produce different networks
P2N can generate 3 kind of networks :
- Simple and mixed networks : simple networks express direct networks; between inventors, applicants ou IPC classes. While mixed networks are intended to describe for example who works for who (inventor-applicant) or an expertise area of an applicant or inventor, from an IPC class - applicant network or an IPC class- inventor network. Within these networks, all nodes are represented by category, highlighted by colors and hyperlinks to WIPO's databases or Espacenet. The nodes and edges of a network are weighed by degree and occurrences to display the cooperation. These weights can be represented graphically with the features of different sizes. Networks can also be used to track the evolution of a patent universe through time.
- Quotes networks : This is the graph of the publications cited by patent, but also with the publications that cite a given patent. In addition, networks can provide a quality assessment based on scientometric evaluation. This assessment can also be based on the network that shows all patents joined at their equivalents.
- Patents Network : This is to provide a hierarchical representation of patent families delivered by the Open Patent Services database.
A user can look at any node by category or URL for the network properties such as degree, centrality and time for better understanding of the patent world with the opportunity to put into practice techniques in the field of network science.
Represent nodes for the different degrees of centrality
An important characteristic of networks is the centrality on patent-inventors-applicants. Centrality is a structural feature of these variables in the network, ie the centrality shows how this variable is in the entire network. Patents with high centrality scores are often the most important or the most cited, they are therefore key information conduits.
The difference between the networks, and similar citations
References means patents that will cite other patents, and external references, such as newspapers, magazines …
- The concept of “outdegree” in this case represents the patents are quite complex and composed of several other patents, ie using patents to improve other. The “outdegree” equivalent are patents networks that have more similar (translations, versions).
- The concept of “Indegree” in turn designates here patents the most importants, key patents. The concept of “Indegree” in quotes'network means patents the most cited ou patents cited by other patents. For example it is the algorithm of Google and PageRank which quantitatively measures the popularity of a website basing on the fact that a site is cited by other sites or other people.