Method and apparatus for generating parallel text in same language

An apparatus for generating a parallel text in same language, comprising:

• at least one processor;
• and
• a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:
• acquiring a source segmented word sequence and a pre-trained word vector table, wherein the pre-trained word vector table characterizes a correlation between a word and a word vector, and a similarity between word vectors of two words having similar semantics is higher than a similarity between word vectors of two words not having similar semantics, the word vector is a real number vector represented by Distributed Representation or Distributional Representation;
• determining a source word vector sequence corresponding to the source segmented word sequence according to the pre-trained word vector table;
• importing the source word vector sequence into a first pre-trained recurrent neural network model, to generate an intermediate vector of a preset dimension for characterizing semantics of the source segmented word sequence, wherein the first recurrent neural network model characterizes a correlation between a word vector sequence and a vector of the preset dimension;
• importing the intermediate vector into a second pre-trained recurrent neural network model, to generate a target word vector sequence corresponding to the intermediate vector, wherein the second recurrent neural network model characterizes a correlation between the vector of the preset dimension and the word vector sequence;
• and
• determining a target segmented word sequence corresponding to the target word vector sequence according to the pre-trained word vector table, and determining the target segmented word sequence as a parallel text in the same language corresponding to the source segmented word sequence, the target segmented word sequence being in the same language and having semantics similar to the source segmented word sequence.

A non-transitory computer readable storage medium storing a computer program, wherein the program, when executed by a processor, implements a method comprising:

• acquiring a source segmented word sequence and a pre-trained word vector table, wherein the pre-trained word vector table characterizes a correlation between a word and a word vector, and a similarity between word vectors of two words having similar semantics is higher than a similarity between word vectors of two words not having similar semantics, the word vector is a real number vector represented by Distributed Representation or Distributional Representation;
• determining a source word vector sequence corresponding to the source segmented word sequence, according to the pre-trained word vector table;
• importing the source word vector sequence into a first pre-trained recurrent neural network model, to generate an intermediate vector of a preset dimension for characterizing semantics of the source segmented word sequence, wherein the first recurrent neural network model characterizes a correlation between a word vector sequence and a vector of the preset dimension;
• importing the intermediate vector into a second pre-trained recurrent neural network model, to generate a target word vector sequence corresponding to the intermediate vector, wherein the second recurrent neural network model characterizes a correlation between the vector of the preset dimension and the word vector sequence;
• and
• determining a target segmented word sequence corresponding to the target word vector sequence according to the pre-trained word vector table, and determining the target segmented word sequence as a parallel text in the same language corresponding to the source segmented word sequence, the target segmented word sequence being in the same language and having semantics similar to the source segmented word sequence.