| 摘要 |
A computerized method for generating and evaluating natural language-generated text involves receiving, in a computer, data input by a user, generating, using a natural language generation technique, multiple instances of text stories based upon both contents of a corpus and the received data; analyzing the multiple instances of text stories as a weighted combination of computed geographic scores, distance scores, information content scores, replacement scores and extra aspect scores, providing a ranked set of the generated text stories to a user, receiving a selection of one of the text stories in the ranked set, and storing the selected story. |
| 主权项 |
1. A computerized method for generating and evaluating natural language-generated text, the method comprising:
receiving, in a computer, data input by a user, the received data to be used to generate multiple different story instances; accessing a corpus stored in non volatile storage, using a processor of the computer; generating, using a natural language generation technique, multiple instances of text stories based upon both contents of the corpus and the received data; analyzing each of the multiple instances of text stories by computing a geographic score using the processor,
computing a distance score using the processor,computing an information content score using the processor,computing a replacement score using the processor, and computing an extra aspect score using the processor; for each of the multiple instances of text stories, normalizing, using the processor, each of the computed geographic score, the computed distance score, the computed information content score, the computed replacement score, and the computed extra aspect score, to a common scale so as to obtain, for each of the multiple instances of text stories, a normalized geographic score, a normalized distance score, a normalized information content score, a normalized replacement score, and a normalized extra aspect score; applying, using the processor, a first weighting factor to all of the normalized geographic scores of each of the multiple instances of text stories to obtain weighted geographic scores; applying, using the processor, a second weighting factor to all of the normalized distance scores of each of the multiple instances of text stories to obtain weighted distance scores; applying, using the processor, a third weighting factor to all of the normalized information content scores of each of the multiple instances of text stories to obtain weighted information content scores; applying, using the processor, a fourth weighting factor to all of the normalized replacement scores of each of the multiple instances of text stories to obtain weighted replacement scores; applying, using the processor, a fifth weighting factor to all of the normalized extra aspect scores of each of the multiple instances of text stories to obtain weighted extra aspect scores; using the processor, computing a total score for each of the multiple instances of text stories from the respective weighted geographic score, weighted distance score, weighted information content score, weighted replacement score, and weighted extra aspect score; generating, for display on a screen associated with the user, an ordered set of the multiple instances of text stories, using the processor, ranked according to their total score; receiving from the user, at the processor, a selection of at least one of the displayed instances of text stories in the ordered set; and creating a file in the non-volatile storage containing the selected at least one of the displayed instances of text stories. |
