Method and apparatus for data structuring of text

1. An apparatus for data structuring of text, the apparatus comprising:

a processor; and

a memory storing instructions executable by the processor,

wherein the processor is configured to execute the instructions to:

extract text and location information of the text from an image based on an optical character recognition (OCR) technique;

generate a text unit based on the text and the location information;

classify a form of the image based on the text;

label the text unit as first text, second text, and third text respectively corresponding to an item name, an item value, and others based on the classified form of the image;

structure the text by mapping the second text corresponding to the item value and the first text corresponding to the item name; and

determine misrecognition of the first text and correct the first text determined to be misrecognized.

2. The apparatus of claim 1 , wherein the processor is configured to classify the form by searching for the text in a search engine, and

the search engine includes a plurality of form samples to label a form most similar to the text.

3. The apparatus of claim 1 , wherein the processor is configured to set text whose distance between the texts is less than or equal to a preset threshold value to one text unit based on the location information of the text.

4. The apparatus of claim 3 , wherein the processor is configured to use the item name corresponding to the classified form in the process of generating the text unit.

5. The apparatus of claim 4 , wherein the processor is configured to apply the text unit to a natural language processing model and labels the text unit as the first text to third text.

6. The apparatus of claim 1 , wherein the processor is configured to identify fourth text, which is the first text that falls within a preset distance threshold value from the second text, and maps the first text corresponding to the fourth text and the second text.

7. The apparatus of claim 6 , wherein when a plurality of fourth texts are identified, the processor is configured to calculate vector similarity of the second text and each of the plurality of fourth texts through a similarity verification model, and

the first text corresponding to a fourth text having highest vector similarity among the plurality of fourth texts is mapped to the second text, and

wherein each of the plurality of fourth texts is the first text falling within the preset distance threshold value from the second text.

8. The apparatus of claim 3 , wherein the processor is configured to calculate similarity of a representative keyword for the item name among training data of the natural language processing model and the first text to determine whether the first text is misrecognized.

9. A method of data structuring of text, the method comprising:

extracting text and location information of the text from an image based on an optical character recognition (OCR) technique;

generating a text unit based on the text and the location information;

classifying a form of the image based on the text;

labeling the text unit as first text, second text, and third text respectively corresponding to an item name, an item value, and others based on the classified form of the image;

structuring the text by mapping the second text corresponding to the item value and the first text corresponding to the item name; and

determining misrecognition of the first text and correcting the first text when the first text is determined to be misrecognized.

10. The method of claim 9 , further comprising identifying fourth text, which is the first text that falls within a preset distance threshold value from the second text, and mapping the first text corresponding to the fourth text and the second text.

11. The method of claim 10 , wherein when a plurality of fourth texts are identified, vector similarity of the second text and each of the plurality of fourth texts is calculated through a similarity verification model, and the first text corresponding to a fourth text having highest vector similarity among the plurality of fourth texts is mapped to the second text, and

wherein each of the plurality of fourth texts is the first falling within the preset distance threshold value from the second text.

12. The method of claim 9 , wherein the labeling the text unit comprising:

adding one of a tag indicating the middle of the item name, a tag indicating the beginning of the item value and a tag indicating the middle of the item value to each of keywords included in the text unit through a natural language processing model;

connecting keywords with related tags; and

classifying each of the connected keywords as one of the first text corresponding to the item name, the second text corresponding to the item value, and the third text corresponding to the others.

13. The apparatus of claim 1 , wherein the processor is configured to:

add one of a tag indicating the middle of the item name, a tag indicating the beginning of the item value and a tag indicating the middle of the item value to each of keywords included in the text unit through natural language processing model,

connect keywords with related tags, and

classify each of the connected keywords as one of the first text corresponding to the item name, the second text corresponding to the item value, and the third text corresponding to the others.