Core Document ProcessingSplitting

Splitter output types

Splitter outputs follow standardized formats. Understanding these formats is essential when working with evaluation sets, webhooks, and API responses.

Split Output Type

Type Definition

1type SplitOutput = {
2  splits: Split[];
3};
4
5type Split = {
6  // Required fields (all present in API response)
7  classificationId: string; // The id of the classification type (set in the splitter config)
8  type: string; // The type of the split document, corresponds to the classificationId
9  startPage: number; // The start page of the split document
10  endPage: number; // The end page of the split document
11  observation: string; // Explanation of the results
12  identifier: string; // Identifier for the split document (e.g. invoice number)
13  fileId: string; // The file ID of the split document
14
15  // Optional field
16  name?: string; // Optional name for the split
17};
18
19// For creating evaluation set items, use this subset:
20type ProvidedSplit = {
21  classificationId: string; // Required
22  type: string; // Required
23  startPage: number; // Required
24  endPage: number; // Required
25};

Example

1{
2  "splits": [
3    {
4      "classificationId": "invoice",
5      "type": "invoice",
6      "startPage": 1,
7      "endPage": 3
8    },
9    {
10      "classificationId": "other",
11      "type": "other",
12      "startPage": 4,
13      "endPage": 5
14    }
15  ]
16}