Permutation invariance for representing linearized tabular data
An embodiment for encoding permutation-invariant representations of linearized tabular data. The embodiment may receive input including tabular data and linearize a column or row within the received tabular data. The embodiment may automatically assign an increasing sequence of position identifiers to each non-delimiting tokenized cell in the linearized column or row until a header delimiter is reached. The embodiment may, in response to reaching the header delimiter, automatically assign a monotonically increasing sequence of position identifiers for each non-delimiting tokenized cell positioned after the header delimiter, restarting from an integer corresponding to 1 greater than the position identifier assigned to the header delimiter for each non-delimiting tokenized cell positioned after cell delimiters. The embodiment may automatically assign a static position identifier for each of the cell delimiters in the linearized column or row and output an encoded permutation-invariant representation of the linearized column or row.
1. A computer-based method of encoding tabular data with permutation invariance, the method comprising:
receiving input including tabular data and linearizing a column or row within the received tabular data;
automatically assigning an increasing sequence of position identifiers to each non-delimiting tokenized cell in the linearized column or row until a header delimiter is reached;
in response to reaching the header delimiter, automatically assigning a monotonically increasing sequence of position identifiers for each non-delimiting tokenized cell positioned after the header delimiter, restarting from an integer corresponding to 1 greater than the position identifier assigned to the header delimiter for each non-delimiting tokenized cell positioned after cell delimiters;
automatically assigning a static position identifier for each of the cell delimiters in the linearized column or row, the static position identifier being 1 greater than a highest position identifier assigned to the non-delimiting tokenized cells; and
automatically outputting an encoded permutation-invariant representation of the linearized column or row.
2. The computer-based method of claim 1 , further comprising inputting the encoded permutation invariant representation into a pretrained transformer model to generate a final vector.
3. The computer-based method of claim 2 , further comprising using the generated final vector as an input for performing table search functions.
4. The computer-based method of claim 2 , further comprising using the final vector as an input for performing for table question-answering functions.
5. The computer-based method of claim 2 , further comprising using the final vector as an input for performing table interpretation functions.
6. The computer-based method of claim 2 , further comprising using the final vector as an input in a down-stream table analytics function carried out in a supervised setup.
7. The computer-based method of claim 2 , further comprising using the final vector as an input for performing a down-stream table analytics function carried out in a zero-shot setting.
8. A computer system, the computer system comprising:
one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage medium, and program instructions stored on at least one of the one or more computer-readable tangible storage medium for execution by at least one of the one or more processors via at least one of the one or more computer-readable memories, wherein the computer system is capable of performing a method comprising:
receiving input including tabular data and linearizing a column or row within the received tabular data;
automatically assigning an increasing sequence of position identifiers to each non-delimiting tokenized cell in the linearized column or row until a header delimiter is reached;
in response to reaching the header delimiter, automatically assigning a monotonically increasing sequence of position identifiers for each non-delimiting tokenized cell positioned after the header delimiter, restarting from an integer corresponding to 1 greater than the position identifier assigned to the header delimiter for each non-delimiting tokenized cell positioned after cell delimiters;
automatically assigning a static position identifier for each of the cell delimiters in the linearized column or row, the static position identifier being 1 greater than a highest position identifier assigned to the non-delimiting tokenized cells; and
automatically outputting an encoded permutation-invariant representation of the linearized column or row.
9. The computer system of claim 8 , further comprising inputting the encoded permutation invariant representation into a pretrained transformer model to generate a final vector.
10. The computer system of claim 9 , further comprising using the generated final vector as an input for performing table search functions.
11. The computer system of claim 9 , further comprising using the final vector as an input for performing for table question-answering functions.
12. The computer system of claim 9 , further comprising using the final vector as an input for performing table interpretation functions.
13. The computer system of claim 9 , further comprising using the final vector as an input in a down-stream table analytics function carried out in a supervised setup.
14. The computer system of claim 9 , further comprising using the final vector as an input for performing a down-stream table analytics function carried out in a zero-shot setting.
15. A computer program product, the computer program product comprising:
one or more computer-readable tangible storage medium and program instructions stored on at least one of the one or more computer-readable tangible storage medium, the program instructions executable by a processor capable of performing a method, the method comprising:
receiving input including tabular data and linearizing a column or row within the received tabular data;
automatically assigning an increasing sequence of position identifiers to each non-delimiting tokenized cell in the linearized column or row until a header delimiter is reached;
in response to reaching the header delimiter, automatically assigning a monotonically increasing sequence of position identifiers for each non-delimiting tokenized cell positioned after the header delimiter, restarting from an integer corresponding to 1 greater than the position identifier assigned to the header delimiter for each non-delimiting tokenized cell positioned after cell delimiters;
automatically assigning a static position identifier for each of the cell delimiters in the linearized column or row, the static position identifier being 1 greater than a highest position identifier assigned to the non-delimiting tokenized cells; and
automatically outputting an encoded permutation-invariant representation of the linearized column or row.
16. The computer program product of claim 15 , further comprising inputting the encoded permutation invariant representation into a pretrained transformer model to generate a final vector.
17. The computer program product of claim 16 , further comprising using the generated final vector as an input for performing table search functions.
18. The computer program product of claim 16 , further comprising using the final vector as an input for performing for table question-answering functions.
19. The computer program product of claim 16 , further comprising using the final vector as an input for performing table interpretation functions.
20. The computer program product of claim 16 , further comprising using the final vector as an input in a down-stream table analytics function carried out in a supervised setup.