CBR4SSPlatform

Setup

This projects uses java-dotenv to load environment variables from a .env file, so you have to make a copy of example.env, name it .env and configure the desire variables

$ cp example.env .env

Variables

Commented variables (i.e. line starts with # ) are set with default value. Uncomment them to set another one.

Database Variables:

DATABASE_HOST: Host of the Database (default 127.0.0.1)
DATABASE_PORT: Port of the Database (default 27017)

DATABASE_KB_NAME: Name of the knowledge base Database (default KB)
DATABASE_QUERIES_NAME: Name of the queries Database (default queries)
DATABASE_RESULTS_PREFIX: Prefix of the name of the results Database (default retrieved_cases)
DATABASE_LEARNED_PREFIX: Prefix where the learned cases will be added (besides of KB) to inspect them (default learned)

Runtime Environment Variables

VERBOSE: Choose if you want to run it in verbose mode (default false)
K: The number of k to use in the knn algorithm (default 10)
SELECTION_RULE: The selection algorithm rule, one of KNN, WKNN and DWKNN (default KNN)
DISTANCE_THRESHOLD: Threshold distance value to retrain the CBR (when is not present, the CBR does not learn) 

Selections Rules

KNN

This selection method is one of the oldest and simplest classifiers. The basic rationale is defined as follows: A query is labelled by a majority vote of its k-nearest neighbors, i.e. the solution of a Case will be the most predominant in its k-nearest neighbors, this is a simple majority vote.

WKNN

Dudani first introduced a weighted voting method for KNN, called the distance-weighted knearest neighbor rule (WKNN). In WKNN, the closer neighbors are weighted more heavily than the farther ones, using the distance-weighted function. The weight w^i for i-th nearest neighbor is defined as follow:

TODO: insert math formula

DWKNN

DWKNN is based on WKNN: to give different weights to k nearest neighbors according to their distances, with closer neighbors having greater weights. Nevertheless, different from the weights in WKNN, we assign to the i-th nearest neighbor xNNi of the query x a dual weight wi, defined by the dual distance-weighted function as below:

TODO: insert math formula