Logtalk 3.100.0 released

[pmoura]

Logtalk 3.100.0 is now available for downloading at:

https://logtalk.org/

This is a major library-focused release centered on a broad expansion of the standard library: machine learning protocols and algorithms, combinatorics and sequence-processing libraries, geospatial/orbital/standards/data-format support, interval reasoning, and hashes/HMAC. It also standardizes classifier naming and export APIs and includes smaller compiler/runtime, documentation, tooling, examples, and IDE improvements.

New libraries by category/area

Machine learning foundations and algorithms

• Category supporting libraries: clustering_protocols, pattern_mining_protocols, frequent_pattern_mining_protocols, sequential_pattern_mining_protocols, regression_protocols, dimension_reduction_protocols, anomaly_detection_protocols, and ranking_protocols
• Support library: linear_algebra for numeric vectors and matrices
• Clustering: agglomerative_clusterer, dbscan_clusterer, gaussian_mixture_clusterer, hdbscan_clusterer, hierarchical_clustering, kcenters_clusterer, kmeans_clusterer, kmedians_clusterer, kmedoids_clusterer, kmodes_clusterer, kprototypes_clusterer, and optics_clusterer
• Pattern mining: apriori_pattern_miner, eclat_pattern_miner, fp_growth_pattern_miner, prefix_span_pattern_miner, clo_span_pattern_miner, gsp_pattern_miner, and spade_pattern_miner
• Regression: bayesian_ridge_regression, gradient_boosting_regression, gaussian_process_regression, knn_regression, linear_regression, ridge_regression, lasso_regression, elastic_net_regression, regression_tree, and random_forest_regression
• Dimension reduction: ica_projection, kernel_pca_projection, lda_projection, nmf_projection, pca_projection, pls_projection, probabilistic_pca_projection, random_projection, and truncated_svd_projection
• Anomaly detection: cusum_anomaly_detector, ewma_anomaly_detector, knn_distance_anomaly_detector, lof_anomaly_detector, z_score_anomaly_detector, modified_z_score_anomaly_detector, and iqr_anomaly_detector
• Ranking: borda_ranker, bradley_terry_ranker, colley_ranker, copeland_ranker, elo_ranker, glicko2_ranker, glicko2_periodic_ranker, hodge_rank, kemeny_young_ranker, massey_ranker, plackett_luce_ranker, plackett_luce_last_ranker, rank_centrality, regularized_bradley_terry_ranker, schulze_ranker, ranked_pairs, and thurstone_mosteller_ranker
• Classification: logistic_regression_classifier, linear_svm_classifier, gradient_boosting_classifier, kernel_svm_classifier, lda_classifier, qda_classifier, and sgd_classifier

Combinatorics / sequences

• arrangements, cartesian_products, derangements, multisets, and partitions

Geospatial

• crs_projections, geojson, nmea, tle_orbits, geohash

Orbital

• ccsds_frames, ccsds_link_profiles, ccsds_packet_services, ccsds_packetization, ccsds_tc_services, ccsds_time_codes, and ccsds_time_fields

Binary / data formats

• ieee_754, hmac, message_pack, json_pointer, and wkt_wkb

Standards / identifiers / finance

• iso_639, iso_3166, iso_4217, iso_9362, and iso_13616

Changed libraries by category/area

Naming / API consistency / migration

• ccsds renamed to ccsds_packets
• classifier_protocols renamed to classification_protocols
• ada_boost, c45, knn, naive_bayes, nearest_centroid, and random_forest renamed to adaptive_boosting_classifier, c45_classifier, knn_classifier, naive_bayes_classifier, nearest_centroid_classifier, and random_forest_classifier
• Classifier export predicates were standardized as export_to_clauses/4 and export_to_file/4

Machine learning infrastructure

• classification_protocols gained check_classifier/1 and valid_classifier/1 plus protected training-dataset validation predicates
• Classifier libraries now support diagnostics and richer metadata exports and share more code through the new classifier_common category
• isolation_forest_anomaly_detector now depends on anomaly_detection_protocols; the functionality is preserved but related export and print predicates were renamed accordingly

Collections / combinatorics / geospatial

• Repetition-allowed ordered selections moved from permutations to arrangements; the former cartesian_product/3 predicate is replaced by arrangements/3-4 and arrangement/3-4
• Derangements moved from permutations to derangements
• Repetition-allowed unordered selections moved from combinations to multisets; the former replacement predicates are replaced by multisets/3-4, multiset/3-4, and count_multisets/3
• combinations and permutations gained new counting, indexing, sampling, distinct-value, and lexicographic stepping predicates
• geospatial now uses geographic(Latitude,Longitude) terms instead of (Latitude,Longitude) pairs and adds new bounding-box predicates

Math / hashing / analytics

• hashes gained keyed digests plus metadata for fixed-size cryptographic hash objects that are safe to use with HMAC
• types gained numberlist::minkowski_distance/4 plus new natural predicates including bell/2, catalan/2, fibonacci/2, lucas/2, multinomial/2, falling_factorial/3, rising_factorial/3, partition_number/2, stirling_first/3, stirling_second/3, and subfactorial/2
• intervals gained Allen interval relations, relation algebra, and symbolic interval constraint network functionality with path-consistency propagation, explanations, and inspection/comparison predicates
• numberlist::softmax/2-3 is now numerically stable

Performance / portability / cleanup

• Tail-recursive rewrites were applied across multiple libraries including amqp, avro, base58, hashes, json_ld, json_schema, graphs, mime_types, toml, toon, tzif, url, yaml, and several classifier libraries
• json_schema now reuses json_pointer, and avro, cbor, and protobuf now reuse ieee_754
• linda, tzif, ulid, url, and yaml now compile all entities in optimal mode; dead code was removed from linda, meta_compiler, sockets, toml, and validations
• numberlist::average/2, median/2, and softmax/2-3 were fixed to work around SWI-Prolog returning integers for float operations

Non-library notable changes

• logtalk_doclet and logtalk_tester: new -z option in the Bash and PowerShell scripts to make a sound on completion
• Compiler/runtime: added new logtalk::halt/0-1 public predicates plus the logtalk::halt_hook/2 user-defined hook predicate
• Documentation: added a Handbook libraries overview grouped by scope and fixed sequence non-terminals in the grammar section
• Examples: added the eclipse_tests_dsl example, sped up the philosophers example tests, and fixed the neo4j example classpath scripts and notes for recent Neo4j releases
• IDE support: added syntax highlighting support for the Fresh editor and fixed an incorrect use_module/1 example in syntax highlighting test files

For details and a complete list of changes, please consult the release notes at:

https://github.com/LogtalkDotOrg/logtalk3/blob/master/RELEASE_NOTES.md