Suggested use

Current implementation of autoBOT enables the user to explore multiple different representation spaces. However, the end-goal is to offer a tool which serves as a very strong baseline. To this end, the configuration below was shown to perform well across multiple benchmarks/shared tasks. Note that this configuration needs the sentence-transformers library (multilingual contextual representations).

import autoBOTLib
import pandas as pd

## Load example data frame
dataframe = pd.read_csv("../data/insults/train.tsv", sep="\t")
train_sequences = dataframe['text_a'].values.tolist()
train_targets = dataframe['label'].values

autoBOTLibObj = autoBOTLib.GAlearner(
        train_sequences,  # input sequences
        train_targets,  # target space
        time_constraint=3,  # time in hours
        num_cpu="all",  # number of CPUs to use
        task_name="example test",  # task identifier
        scoring_metric = "f1", # sklearn-compatible scoring metric as the fitness.
        hof_size=3,  # size of the hall of fame
        top_k_importances=25,  # how many top features to output as final ranking
        memory_storage=
        "./memory",  # triplet base for concept features (see ./examples folder)
        representation_type="neurosymbolic")

autoBOTLibObj.evolve(
                nind=10,  ## population size
                crossover_proba=0.6,  ## crossover rate
                mutpb=0.4)  ## mutation rate

The triplet knowledge bases can be downloaded from e.g., https://github.com/totogo/awesome-knowledge-graph#knowledge-graph-dataset. See the autobot/examples for more examples.