@mat3ra/esse 2025.8.14-0 → 2025.10.8-0

package/example/software_directory/ml/unit/processing/data_transformation/manipulation.json

@@ -1,14 +0,0 @@
-{
-    "flowchartId": "data_transformation",
-    "head": true,
-    "inputData": {
-        "cleanMissingData": true,
-        "removeDuplicateRows": true,
-        "replaceNoneValuesWith": 0
-    },
-    "name": "data_transformation",
-    "operation": "data_transformation",
-    "operationType": "manipulation",
-    "status": "idle",
-    "type": "processing"
-}

package/example/software_directory/ml/unit/processing/data_transformation/scale_and_reduce.json

@@ -1,26 +0,0 @@
-{
-    "flowchartId": "data_transformation",
-    "head": true,
-    "inputData": {
-        "perFeature": [
-            {
-                "mean": 0.156,
-                "name": "atomic_radius:Ge",
-                "scale": 1.5,
-                "variance": 0.1
-            },
-            {
-                "mean": 0.155,
-                "name": "atomic_radius:Si",
-                "scale": 1.4,
-                "variance": 0.11
-            }
-        ],
-        "scaler": "standard_scaler"
-    },
-    "name": "data_transformation",
-    "operation": "data_transformation",
-    "operationType": "scale_and_reduce",
-    "status": "idle",
-    "type": "processing"
-}

package/example/software_directory/ml/unit/processing/feature_selection/filter_based.json

@@ -1,13 +0,0 @@
-{
-    "flowchartId": "feature_selection",
-    "head": true,
-    "inputData": {
-        "algorithm": "f_regression",
-        "nFeatures": 100
-    },
-    "name": "feature_selection",
-    "operation": "feature_selection",
-    "operationType": "filter_based",
-    "status": "idle",
-    "type": "processing"
-}

package/schema/property/base.json

@@ -1,58 +0,0 @@
-{
-    "$id": "property/base",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "Schema of base material's preliminary property",
-    "type": "object",
-    "allOf": [
-        {
-            "description": "entity schema",
-            "$ref": "../in_memory_entity/base.json"
-        }
-    ],
-    "properties": {
-        "slug": {
-            "description": "property slug, e.g. total_energy",
-            "type": "string"
-        },
-        "group": {
-            "description": "property group, e.g. qe:dft:gga:pbe",
-            "type": "string"
-        },
-        "data": {
-            "description": "container of the information, specific to each property",
-            "type": "object"
-        },
-        "source": {
-            "$ref": "source.json"
-        },
-        "exabyteId": {
-            "description": "Id of the corresponding item in the entity bank that this property is obtained for",
-            "type": "array",
-            "items": {
-                "type": "string"
-            }
-        },
-        "precision": {
-            "type": "object"
-        },
-        "count": {
-            "description": "total number of properties among which this property is the best.",
-            "type": "number"
-        },
-        "systemTags": {
-            "description": "property system tags, marks property system characteristics, values refined or best (could be both)",
-            "type": "array",
-            "items": {
-                "type": "string",
-                "enum": [
-                    "isRefined",
-                    "isBest"
-                ]
-            }
-        }
-    },
-    "required": [
-        "data",
-        "source"
-    ]
-}

package/schema/property/meta.json

@@ -1,11 +0,0 @@
-{
-    "$id": "property/meta",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "Schema of material's meta properties",
-    "type": "object",
-    "allOf": [
-        {
-            "$ref": "raw.json"
-        }
-    ]
-}

package/schema/property/raw.json

@@ -1,58 +0,0 @@
-{
-    "$id": "property/raw",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "Schema of material's preliminary property",
-    "type": "object",
-    "allOf": [
-        {
-            "description": "entity schema",
-            "$ref": "../in_memory_entity/base.json"
-        }
-    ],
-    "properties": {
-        "slug": {
-            "description": "property slug, e.g. total_energy",
-            "type": "string"
-        },
-        "group": {
-            "description": "property group, e.g. qe:dft:gga:pbe",
-            "type": "string"
-        },
-        "data": {
-            "description": "container of the information, specific to each property",
-            "type": "object"
-        },
-        "source": {
-            "$ref": "source.json"
-        },
-        "exabyteId": {
-            "description": "Id of the corresponding item in the entity bank that this property is obtained for",
-            "type": "array",
-            "items": {
-                "type": "string"
-            }
-        },
-        "precision": {
-            "type": "object"
-        },
-        "count": {
-            "description": "total number of properties among which this property is the best.",
-            "type": "number"
-        },
-        "systemTags": {
-            "description": "property system tags, marks property system characteristics, values refined or best (could be both)",
-            "type": "array",
-            "items": {
-                "type": "string",
-                "enum": [
-                    "isRefined",
-                    "isBest"
-                ]
-            }
-        }
-    },
-    "required": [
-        "data",
-        "source"
-    ]
-}

package/schema/software_directory/ml/exabyteml.json

@@ -1,23 +0,0 @@
-{
-    "$id": "software-directory/ml/exabyteml",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "exabyte machine learning engine schema",
-    "type": "object",
-    "properties": {
-        "name": {
-            "enum": [
-                "exabyteml"
-            ]
-        },
-        "summary": {
-            "enum": [
-                "exabyte machine learning engine"
-            ]
-        },
-        "version": {
-            "enum": [
-                "0.2.0"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/execution/evaluate/cross-validate.json

@@ -1,26 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/execution/evaluate/cross-validate",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "cross-validation unit schema",
-    "type": "object",
-    "allOf": [
-        {
-            "$ref": "../../../../../workflow/unit/execution.json"
-        }
-    ],
-    "properties": {
-        "input": {
-            "description": "TODO: consider keeping executable `evaluate` and flavor `cross-validate` as before",
-            "type": "object",
-            "properties": {
-                "nSplits": {
-                    "description": "number of groups to split the training dataset for cross-validation",
-                    "type": "number"
-                }
-            },
-            "required": [
-                "nSplits"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/execution/initialize.json

@@ -1,30 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/execution/initialize",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "initialize unit schema",
-    "type": "object",
-    "allOf": [
-        {
-            "$ref": "../../../../workflow/unit/execution.json"
-        }
-    ],
-    "properties": {
-        "input": {
-            "description": "model init unit (NOTE: info about method, eg. regression/linear is taken from (sub)workflow)",
-            "type": "object",
-            "properties": {
-                "targets": {
-                    "description": "target properties to predict (NOTE: must be a subset of targets for which training was done)",
-                    "type": "array",
-                    "items": {
-                        "description": "material features (properties) in a 'flattened' format",
-                        "type": "string"
-                    }
-                }
-            },
-            "required": [
-                "targets"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/execution/score.json

@@ -1,14 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/execution/score",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "train score schema",
-    "type": "object",
-    "allOf": [
-        {
-            "$ref": "../../../../workflow/unit/execution.json"
-        }
-    ],
-    "properties": {
-        "input": {}
-    }
-}

package/schema/software_directory/ml/unit/execution/train.json

@@ -1,39 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/execution/train",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "train unit schema",
-    "type": "object",
-    "allOf": [
-        {
-            "$ref": "../../../../workflow/unit/execution.json"
-        }
-    ],
-    "properties": {
-        "input": {
-            "description": "model train unit (NOTE: info about method, eg. regression/linear is taken from (sub)workflow)",
-            "type": "object",
-            "properties": {
-                "features": {
-                    "description": "material features used for model fitting",
-                    "type": "array",
-                    "items": {
-                        "description": "material features (properties) in a 'flattened' format",
-                        "type": "string"
-                    }
-                },
-                "targets": {
-                    "description": "target properties to train for",
-                    "type": "array",
-                    "items": {
-                        "description": "material features (properties) in a 'flattened' format",
-                        "type": "string"
-                    }
-                }
-            },
-            "required": [
-                "features",
-                "targets"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/execution.json

@@ -1,17 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/execution",
-    "oneOf": [
-        {
-            "$ref": "execution/evaluate/cross-validate.json"
-        },
-        {
-            "$ref": "execution/train.json"
-        },
-        {
-            "$ref": "execution/score.json"
-        },
-        {
-            "$ref": "execution/initialize.json"
-        }
-    ]
-}

package/schema/software_directory/ml/unit/processing/data_transformation/manipulation.json

@@ -1,50 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing/data-transformation/manipulation",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "manipulation unit schema",
-    "allOf": [
-        {
-            "$ref": "../../../../../workflow/unit/processing.json"
-        }
-    ],
-    "type": "object",
-    "properties": {
-        "operation": {
-            "type": "string",
-            "enum": [
-                "data_transformation"
-            ]
-        },
-        "operationType": {
-            "type": "string",
-            "enum": [
-                "manipulation"
-            ]
-        },
-        "inputData": {
-            "type": "object",
-            "properties": {
-                "cleanMissingData": {
-                    "description": "whether to clean missing data, eg. NaN",
-                    "type": "boolean",
-                    "default": true
-                },
-                "removeDuplicateRows": {
-                    "description": "whether to remove duplicate rows",
-                    "type": "boolean",
-                    "default": true
-                },
-                "replaceNoneValuesWith": {
-                    "description": "replace None values with a given value",
-                    "default": 0,
-                    "type": "number"
-                }
-            },
-            "required": [
-                "cleanMissingData",
-                "removeDuplicateRows",
-                "replaceNoneValuesWith"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/processing/data_transformation/scale_and_reduce.json

@@ -1,70 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing/data-transformation/scale-and-reduce",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "scale and reduce unit schema",
-    "allOf": [
-        {
-            "$ref": "../../../../../workflow/unit/processing.json"
-        }
-    ],
-    "type": "object",
-    "properties": {
-        "operation": {
-            "type": "string",
-            "enum": [
-                "data_transformation"
-            ]
-        },
-        "operationType": {
-            "type": "string",
-            "enum": [
-                "scale_and_reduce"
-            ]
-        },
-        "inputData": {
-            "type": "object",
-            "properties": {
-                "scaler": {
-                    "description": "type of scaler to be applied",
-                    "type": "string",
-                    "default": "standard_scaler",
-                    "enum": [
-                        "standard_scaler"
-                    ]
-                },
-                "perFeature": {
-                    "description": "per-feature scaling data",
-                    "type": "array",
-                    "items": {
-                        "type": "object",
-                        "properties": {
-                            "variance": {
-                                "description": "variance in original training data",
-                                "type": "number"
-                            },
-                            "mean": {
-                                "description": "mean value of the original training data",
-                                "type": "number"
-                            },
-                            "scale": {
-                                "description": "scale multiplier for this feature/property",
-                                "type": "number"
-                            },
-                            "name": {
-                                "description": "feature/property name in 'flattened' format",
-                                "type": "string"
-                            }
-                        },
-                        "required": [
-                            "name",
-                            "scale"
-                        ]
-                    }
-                }
-            },
-            "required": [
-                "scaler"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/processing/data_transformation.json

@@ -1,8 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing/data-transformation",
-    "oneOf": [
-        {
-            "$ref": "data_transformation/scale_and_reduce.json"
-        }
-    ]
-}

package/schema/software_directory/ml/unit/processing/feature_selection/filter_based.json

@@ -1,45 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing/feature-selection/filter-based",
-    "$schema": "http://json-schema.org/draft-07/schema#",
-    "title": "filter-based feature selection unit schema",
-    "allOf": [
-        {
-            "$ref": "../../../../../workflow/unit/processing.json"
-        }
-    ],
-    "type": "object",
-    "properties": {
-        "operation": {
-            "type": "string",
-            "enum": [
-                "feature_selection"
-            ]
-        },
-        "operationType": {
-            "type": "string",
-            "enum": [
-                "filter_based"
-            ]
-        },
-        "inputData": {
-            "type": "object",
-            "properties": {
-                "nFeatures": {
-                    "description": "number of features to select for model training. If equal to 0, will use all available features",
-                    "type": "number"
-                },
-                "algorithm": {
-                    "description": "feature selection algorithm following sklearn.feature_selection",
-                    "type": "string",
-                    "enum": [
-                        "f_regression"
-                    ]
-                }
-            },
-            "required": [
-                "nFeatures",
-                "algorithm"
-            ]
-        }
-    }
-}

package/schema/software_directory/ml/unit/processing/feature_selection.json

@@ -1,8 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing/feature-selection",
-    "oneOf": [
-        {
-            "$ref": "feature_selection/filter_based.json"
-        }
-    ]
-}

package/schema/software_directory/ml/unit/processing.json

@@ -1,11 +0,0 @@
-{
-    "$id": "software-directory/ml/unit/processing",
-    "oneOf": [
-        {
-            "$ref": "processing/data_transformation.json"
-        },
-        {
-            "$ref": "processing/feature_selection.json"
-        }
-    ]
-}