Papyrus is a software too Predictive made of a layered system, and it is a network service for data-intensive computing and distributed data mining. There are four layers, and Papyrus can be designed to access any one of them as per our requirements. The applicants are restricted
to move data(MD) from node to node by using the lower layer or application that can move models(MM) or results (MR) from node to node using the upper layer or the top layer.
The data management layer-Osiris
The lower level is the data management layer specially designed to support the critical data mining of clusters, mega-clusters, and super-clusters. It can also be called a global data warehouse. Osiris is divided into folios, and further, these folios are divided into units called segments. Segments are used to move from node to node when requiring for computation. It is only feasible to move elements within super-clusters and clusters that are using broadband and have high-performance networks.
Data mining layer-Anubis
The data mining layer can be viewed as extracting the information from the learning set and data warehouse and the semi-automatic production using the appropriate data mining algorithm of a rule set or predictive model. These predictive models and rules have developed an XML mark-up language called Predictive Model Markup Language(PMML). Osiris manages the input of the data mining layer and the output in the PMML files, governed by the predictive modeling layer called Anubis. This layer consists of a unique application called a specific code. Few of the applications are using the little modified form of C4.5.
A predictive modeling layer
To manage the predictive models, they plan to employ the layers made significantly above the data mining layer. Papyrus is not using the layer, or currently, it does not have one but, instead, uses an agent layer described for managing predictive models.
An agent layer called Bast
Identifying relevant clusters within the mega-cluster or super-cluster, relevant data sets within the collections, appropriate strategies for moving data, models, results, and relevant attributes for particular queries is the role of this layer. Bast will get all the information about the clusters
and the information and data about the collections. This information contains there examining local files which express the information using an XML language called Data Space Markup Language(DSML).