New patent application published for: SEARCH SYSTEM AND SEARCH METHOD

SEARCH SYSTEM AND SEARCH METHOD

Claims

1. A search system comprising: a processor, the processor configured to: extract at least one text data from input document data, the at least one text data including an image label which is assigned to image data; obtain a tag from the at least one text data, wherein the tag contains at least one word related to the image label, and wherein the tag is obtained based on a reference numeral attached to the word; calculate similarities between the tag and a plurality of database tags stored in a database, wherein each of the plurality of database tags is associated with database image data of database document data, and wherein each of the plurality of database tags includes at least one word which relates to an image label assigned to one of the database image data; and identify at least one of the database image data based on the similarities.
2. The search system according to claim 1, wherein the processor is further configured to extract, as first database text, at least a part of paragraphs including an image label from paragraphs included in database document data stored in the database, and use the first database text for obtaining the plurality of database tags.
3. The search system according to claim 1, wherein the processor is further configured to extract, as first text, at least a part of paragraphs including an image label from paragraphs included in the input document data, and use the first text for obtaining the tag.
4. The search system according to claim 1, wherein the at least one text data is extracted from the input document data with a machine learning model on the basis of the image data and the document data, and wherein the machine learning model is learned using learning image data to which a learning image label is assigned and learning document data including the learning image label.
5. The search system according to claim 1, wherein the similarities are calculated using at least one of Jaccard index, Dice index and Simpson index.
6. The search system according to claim 1, wherein the calculation of the similarities uses vectorized data of the words contained in the tag and the plurality of the database tags.
7. The search system according to claim 1, wherein the reference numeral is at least one of an alphabet, a Greek character, a Japanese phonetic alphabet, and a number.
8. The search system according to claim 1, wherein the image label included in the at least one text data is a figure number.
9. A search method comprising: extracting at least one text data from input document data, the at least one text data including an image label which is assigned to image data; obtaining a tag from the at least one text data, wherein the tag contains at least one word related to the image label, and wherein the tag is obtained based on a reference numeral attached to the word; calculating similarities between the tag and a plurality of database tags stored in a database, wherein each of the plurality of database tags is associated with database image data of database document data, and wherein each of the plurality of database tags includes at least one word which relates to an image label assigned to one of the database image data; and identifying at least one of the database image data based on the similarities.
10. The search method according to claim 9, wherein the step of extracting the at least one text data comprises: extracting, as first database text, at least a part of paragraphs including an image label from paragraphs included in database document data stored in the database.
11. The search method according to claim 9, wherein the step of extracting the at least one text data comprises: extracting, as first text, at least a part of paragraphs including an image label from paragraphs included in the input document data.
12. The search method according to claim 9, wherein the at least one text data is extracted from the input document data with a machine learning model on the basis of the image label and the document data, and wherein the machine learning model is learned using a learning image label and learning document data including the learning image label.
13. The search method according to claim 9, wherein the similarities are calculated using at least one of Jaccard index, Dice index and Simpson index.
14. The search method according to claim 9, wherein the calculation of the similarities uses vectorized data of the words contained in the tag and the plurality of the database tags.
15. The search method according to claim 9, wherein the reference numeral is at least one of an alphabet, a Greek character, a Japanese phonetic alphabet, and a number.
16. The search method according to claim 9, wherein the image label included in the at least one text data is a figure number.