Correction of user input
Systems and methods for receiving a first string from a user in a user session, tracking a first response to the first string from the user in the user session, receiving, after receiving the first string, a second string from the user in the user session, the second string having a character operator difference that is below a threshold number from the first string, tracking a second response to the second string from the user in the user session, associating the first string with the second string in response to the second response comprising a higher number of responses as compared with the first response, removing the association between the first string with the second string in response to the second string being a correction of the first string, and correcting, using the association, user input by replacing a portion of the user input that matches the first string with the second string.
1. A method for correction of user input to a user session hosted by a computing system, the method comprising:
receiving a first string from a user in a first user session;
tracking a first response to the first string from the user in the first user session;
receiving, after receiving the first string, a second string from the user in the first user session, the second string having a character operator difference that is below a threshold number from the first string;
tracking a second response to the second string from the user in the first user session;
associating the first string with the second string in response to the second response comprising a higher number of responses as compared with the first response;
removing the association between the first string with the second string in response to the second string being a correction of the first string; and
correcting, using the association, user input to a second user session by replacing a portion of the user input that matches the first string with the second string.
2. The method of claim 1 , wherein the first string and the second string are selected using bigram character analysis.
3. The method of claim 1 , further comprising removing the association in response to a language model use value for the first string in the string pair being above a threshold value.
4. The method of claim 1 , further comprising removing the association in response to a language model use ratio being above a threshold value, the language model use ratio being a ratio of a language model use value of the first string to a language model use value of the second string in the string pair to be removed.
5. The method of claim 1 , further comprising removing the association based on the second string in the selected string pair being a refinement of the first string in the selected string pair.
6. The method of claim 1 , further comprising removing the association based on either the first string of the selected string pair or the second string in the selected string pair being in a predetermined language.
7. The method of claim 1 , wherein the character operator difference is based on Damereau-Levenshtein distance.
8. A computer system to correct user input to a user session hosted by the computing system, the system comprising:
a hardware processor;
a memory device storing instructions that, when performed by the hardware processor, cause the computer system to perform operations comprising:
receiving a first string in a first user session;
tracking a first response to the first string from the user in the first user session;
receiving, after receiving the first string, a second string from the user in the first user session, the second string having a character operator difference that is below a threshold number from the first string;
tracking a second response to the second string from the user in the first user session;
associating the first string with the second string in response to the second response comprising a higher number of responses as compared with the first response;
removing an association between the first string with the second string in response to the second string being a correction of the first string; and
correcting, using the association, user input to a second user session by replacing a portion of the user input that matches the first string with the second string.
9. The computer system of claim 8 , wherein the first string and the second string are selected using bigram character analysis.
10. The computer system of claim 8 , wherein the operations further comprise removing the association in response to a language model use value for the first string in the string pair being above a threshold value.
11. The computer system of claim 8 , wherein the operations further comprise removing the association in response to a language model use ratio being above a threshold value, the language model use ratio being a ratio of a language model use value of the first string to a language model use value of the second string in the string pair to be removed.
12. The computer system of claim 8 , wherein the operations further comprise removing the association based on the second string in the selected string pair being a refinement of the first string in the selected string pair.
13. The computer system of claim 8 , wherein the operations further comprise removing the association based on either the first string of the selected string pair or the second string in the selected string pair being in a predetermined language.
14. The computer system of claim 8 , wherein the character operator difference is based on Damereau-Levenshtein distance.
15. A non-transitory machine-readable hardware medium having instructions stored thereon, which, when executed by a processor, cause the processor to perform operations to correct user input to a user session hosted by a computing system, the operations comprising:
receiving a first string from a user in a first user session;
tracking a first response to the first string from the user in the first user session;
receiving, after receiving the first string, a second string from the user in the first user session, the second string having a character operator difference that is below a threshold number from the first string;
tracking a second response to the second string from the user in the first user session;
associating the first string with the second string in response to the second response comprising a higher number of responses as compared with the first response;
removing an association between the first string with the second string in response to the second string being a correction of the first string; and
correcting, using the association, user input to a second user session by replacing a portion of the user input that matches the first string with the second string.
16. The non-transitory machine-readable hardware medium of claim 15 , wherein the first string and the second string are selected using bigram character analysis.
17. The non-transitory machine-readable hardware medium of claim 15 , wherein the operations further comprise removing the association in response to a language model use value for the first string in the string pair being above a threshold value.
18. The non-transitory machine-readable hardware medium of claim 15 , wherein the operations further comprise removing the association in response to a language model use ratio being above a threshold value, the language model use ratio being a ratio of a language model use value of the first string to a language model use value of the second string in the string pair to be removed.
19. The non-transitory machine-readable hardware medium of claim 15 ; wherein the operations further comprise removing the association based on the second string in the selected string pair being a refinement of the first string in the selected string pair.
20. The non-transitory machine-readable hardware medium of claim 15 , wherein the operations further comprise removing the association based on either the first string of the selected string pair or the second string in the selected string pair being in a predetermined language.