"frbsJpmml", an application written in Java, is used as predictor engine for fuzzy rule-based systems (FRBS) models that are specified in frbsPMML. frbsPMML is a universal framework representation for FRBS which is adopted from the Predictive Model Markup Language (PMML) format.
PMML is an open standard dedicated to the representation of data mining models and sharing of dierent applications. It is developed by the Data Mining Group (DMG) at http://www.dmg.org, and based on the Extensible Markup Language (XML). Basically, PMML contains the following components:
- Header: It contains general information about PMML document.
- DataDictionary: It contains information related to fields or variables.
- MODEL-ELEMENT: It is a main part of the PMML document that consists of models supported by
Figure below shows a PMML workflow, together with some advantages of PMML in data analysis processes. It generally involves a modeling, expert interention, and deploying phase. Moreover, it can be seen that the PMML representation makes a model more interpretable so that human experts can modify and visualize it.
Regarding an FRBS model in frbsPMML, it involves three important elements:
- InferenceSchema: It represents parameters required to do reasoning, such as conjunction and disjunction operators.
- Database: It represents a type of membership functions (MF), parameters of MF, definitions of linguistic values.
- Rulebase: It contains a set of rules.
Furthermore, we provide three different models: Mamdani, Takagi Sugeno Kang, and fuzzy rule-based classification systems. Other useful features to simplify the model are considered, e.g., linguistic hedges, the "don_t care" value, and negation.
It should be noted that "frbsJpmml" is only used for importing PMML format and doing prediction over new data. Therefore, to generate an FRBS model we provide another application which is "frbs". It is written in R and used for constructing an FRBS model by learning from data and from human experts. The detailed information regarding "frbs" can be found in CRAN at http://cran.r-project.org/package=frbs. Additional information about "frbs" is also provided in this web site (http://dicits.ugr.es/software/FRBS/). However, to give better explanations in this site, we provide some examples involving both applications, "frbsJpmml" and "frbs".
The following is a figure showing a workflow and interactions between "frbs" and "frbsJpmml." It has been shown that we can produce frbsPMML by using "frbs", and then we import the representation by calling a function in "frbs" and "frbsJpmml."