Session information processing method and device and storage medium

1. A session information processing method performed at a computing device having one or more processors and memory storing programs to be executed by the one or more processors, the method comprising:

extracting a to-be-analyzed sentence and a preset quantity of preceding sentences of the to-be-analyzed sentence from a session;

performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences, to obtain a first feature set comprising a plurality of first features;

extracting a second feature set comprising one or more second features from a first word set corresponding to the to-be-analyzed sentence and a second word set corresponding to the preset quantity of preceding sentences, one second feature comprising a phrase or sentence comprising a first word and a second word, the first word being one or more words in the first word set, and the second word being one or more words in the second word set; and

determining, according to the first feature set and the second feature set, a sentence category to which the to-be-analyzed sentence belongs, the sentence category comprising a first category indicating that a sentence is complete, and semantics is unambiguous and a second category indicating that a sentences is incomplete, or semantics is ambiguous.

2. The method according to claim 1 , wherein the performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences comprises:

splicing the to-be-analyzed sentence and the preset quantity of preceding sentences according to a sentence sequence in the session, to obtain a spliced sentence, adjacent preceding sentences being separated by using a first character, and the to-be-analyzed sentence and a previous sentence adjacent to the to-be-analyzed sentence being separated by using a second character; and

extracting an N-gram feature from the spliced sentence and using the N-gram feature as the first feature to form the first feature set, N being a preset integer.

3. The method according to claim 1 , wherein the determining, according to the first feature set and the second feature set, a sentence category to which the to-be-analyzed sentence belongs comprises:

encoding each feature in the first feature set and the second feature set, to obtain a first vector corresponding to the feature, a quantity of elements in each first vector being the same;

determining a second vector according to the first vectors, the second vector being a vector representing the to-be-analyzed sentence and the preset quantity of preceding sentences;

inputting the second vector into a preset classifier, to obtain matching degrees between the to-be-analyzed sentence and sentence categories; and

determining, according to the matching degrees between the to-be-analyzed sentence and the sentence categories, the sentence category to which the to-be-analyzed sentence belongs.

4. The method according to claim 3 , wherein before the encoding each feature in the first feature set and the second feature set, the method further comprises:

inputting each first feature into a preset hash function, to obtain a hash value corresponding to the first feature, the hash function being capable of mapping an input feature to an integer in a preset interval; and

the encoding each feature in the first feature set and the second feature set, to obtain a first vector of a preset dimension corresponding to the feature comprises:

encoding first features having a same hash value as a feature, to obtain the corresponding first vector.

5. The method according to claim 4 , wherein the hash function comprises:

f 1 ( x 1 )= x 1 mod n , wherein

x1 is an input feature of the hash function, and f1(x1) is a hash value and is an integer in [0, n−1].

6. The method according to claim 3 , wherein the preset quantity is a first preset quantity, and before the inputting the second vector into a preset classifier, the method further comprises:

inputting the second vector into a transformation model, the transformation model comprising a second preset quantity of transformation functions, nonlinear transformation being capable of being performed on input data by using the transformation functions, and in two adjacent transformation functions, output data of a former transformation function being input data of a latter transformation function in a case that the second preset quantity is greater than or equal to 2; and

the inputting the second vector into a preset classifier comprises:

inputting output data of the transformation model into the preset classifier.

7. The method according to claim 6 , wherein an ith transformation function is:

h i =f 2 ( W i *h i-1 +b i ), wherein

f2(x2)=1/(1+exp(−x2)), x2=W i *h i-1 +b i , Wi is a weight coefficient, bi is a bias coefficient, hi−1 is output data of an (i−1)th transformation function, and h0 is the second vector when i is 1.

8. The method according to claim 3 , wherein the determining a second vector according to the first vectors comprises:

determining an average vector of the first vectors, and using the average vector as the second vector, each element in the average vector being a ratio of a sum of elements at corresponding locations in the first vectors to a quantity of the first vectors.

9. The method according to claim 3 , wherein the preset classifier calculates the matching degrees between the to-be-analyzed sentence and the sentence categories by using the following formula:

soft

⁢

⁢

max

⁡

(

y

0

,

…

⁢

,

y

n

)

=

(

y

0

∑

j

=

0

n

⁢

y

j

,

…

⁢

,

y

n

∑

j

=

0

n

⁢

y

j

)

,

wherein

y 0 , . . . , y n are elements in vectors inputted into the preset classifier.

10. The method according to claim 3 , wherein the determining, according to the matching degrees between the to-be-analyzed sentence and the sentence categories, the sentence category to which the to-be-analyzed sentence belongs comprises:

using a sentence category corresponding to a highest matching degree in the matching degrees between the to-be-analyzed sentence and the sentence categories as the sentence category to which the to-be-analyzed sentence belongs.

11. A computing device comprising one or more processors, memory coupled to the one or more processors and a plurality of programs stored in the memory that, when executed by the one or more processors, cause the computing device to perform a plurality of operations comprising:

extracting a to-be-analyzed sentence and a preset quantity of preceding sentences of the to-be-analyzed sentence from a session;

performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences, to obtain a first feature set comprising a plurality of first features;

extracting a second feature set comprising one or more second features from a first word set corresponding to the to-be-analyzed sentence and a second word set corresponding to the preset quantity of preceding sentences, one second feature comprising a phrase or sentence comprising a first word and a second word, the first word being one or more words in the first word set, and the second word being one or more words in the second word set; and

determining, according to the first feature set and the second feature set, a sentence category to which the to-be-analyzed sentence belongs, the sentence category comprising a first category indicating that a sentence is complete, and semantics is unambiguous and a second category indicating that a sentences is incomplete, or semantics is ambiguous.

12. The computing device according to claim 11 , wherein the performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences comprises:

splicing the to-be-analyzed sentence and the preset quantity of preceding sentences according to a sentence sequence in the session, to obtain a spliced sentence, adjacent preceding sentences being separated by using a first character, and the to-be-analyzed sentence and a previous sentence adjacent to the to-be-analyzed sentence being separated by using a second character; and

extracting an N-gram feature from the spliced sentence and using the N-gram feature as the first feature to form the first feature set, N being a preset integer.

13. The computing device according to claim 11 , wherein the determining, according to the first feature set and the second feature set, a sentence category to which the to-be-analyzed sentence belongs comprises:

encoding each feature in the first feature set and the second feature set, to obtain a first vector corresponding to the feature, a quantity of elements in each first vector being the same;

determining a second vector according to the first vectors, the second vector being a vector representing the to-be-analyzed sentence and the preset quantity of preceding sentences;

inputting the second vector into a preset classifier, to obtain matching degrees between the to-be-analyzed sentence and sentence categories; and

determining, according to the matching degrees between the to-be-analyzed sentence and the sentence categories, the sentence category to which the to-be-analyzed sentence belongs.

14. The computing device according to claim 13 , wherein before the encoding each feature in the first feature set and the second feature set, the method further comprises:

inputting each first feature into a preset hash function, to obtain a hash value corresponding to the first feature, the hash function being capable of mapping an input feature to an integer in a preset interval; and

the encoding each feature in the first feature set and the second feature set, to obtain a first vector of a preset dimension corresponding to the feature comprises:

encoding first features having a same hash value as a feature, to obtain the corresponding first vector.

15. The computing device according to claim 13 , wherein the preset quantity is a first preset quantity, and before the inputting the second vector into a preset classifier, the method further comprises:

inputting the second vector into a transformation model, the transformation model comprising a second preset quantity of transformation functions, nonlinear transformation being capable of being performed on input data by using the transformation functions, and in two adjacent transformation functions, output data of a former transformation function being input data of a latter transformation function in a case that the second preset quantity is greater than or equal to 2; and

the inputting the second vector into a preset classifier comprises:

inputting output data of the transformation model into the preset classifier.

16. The computing device according to claim 13 , wherein the determining a second vector according to the first vectors comprises:

determining an average vector of the first vectors, and using the average vector as the second vector, each element in the average vector being a ratio of a sum of elements at corresponding locations in the first vectors to a quantity of the first vectors.

17. The computing device according to claim 13 , wherein the preset classifier calculates the matching degrees between the to-be-analyzed sentence and the sentence categories by using the following formula:

soft

⁢

⁢

max

⁡

(

y

0

,

…

⁢

,

y

n

)

=

(

y

0

∑

j

=

0

n

⁢

y

j

,

…

⁢

,

y

n

∑

j

=

0

n

⁢

y

j

)

,

wherein

y 0 , . . . , y n are elements in vectors inputted into the preset classifier.

18. The computing device according to claim 13 , wherein the determining, according to the matching degrees between the to-be-analyzed sentence and the sentence categories, the sentence category to which the to-be-analyzed sentence belongs comprises:

using a sentence category corresponding to a highest matching degree in the matching degrees between the to-be-analyzed sentence and the sentence categories as the sentence category to which the to-be-analyzed sentence belongs.

19. A non-transitory computer readable storage medium storing a plurality of machine readable instructions in connection with a computing device having one or more processors, wherein the plurality of machine readable instructions, when executed by the one or more processors, cause the computing device to perform a plurality of operations including:

extracting a to-be-analyzed sentence and a preset quantity of preceding sentences of the to-be-analyzed sentence from a session;

performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences, to obtain a first feature set comprising a plurality of first features;

extracting a second feature set comprising one or more second features from a first word set corresponding to the to-be-analyzed sentence and a second word set corresponding to the preset quantity of preceding sentences, one second feature comprising a phrase or sentence comprising a first word and a second word, the first word being one or more words in the first word set, and the second word being one or more words in the second word set; and

determining, according to the first feature set and the second feature set, a sentence category to which the to-be-analyzed sentence belongs, the sentence category comprising a first category indicating that a sentence is complete, and semantics is unambiguous and a second category indicating that a sentences is incomplete, or semantics is ambiguous.

20. The non-transitory computer readable storage medium according to claim 19 , wherein the performing word segmentation on the to-be-analyzed sentence and the preset quantity of preceding sentences comprises:

splicing the to-be-analyzed sentence and the preset quantity of preceding sentences according to a sentence sequence in the session, to obtain a spliced sentence, adjacent preceding sentences being separated by using a first character, and the to-be-analyzed sentence and a previous sentence adjacent to the to-be-analyzed sentence being separated by using a second character; and

extracting an N-gram feature from the spliced sentence and using the N-gram feature as the first feature to form the first feature set, N being a preset integer.