Learning method and generating apparatus
AssigneeFujitsu Limited
InventorTakuya Makino
Life of this patent
Estimating the term…
Abstract
A computer-implemented learning method includes combining a first score for a word included in a dictionary of a model and a second score for the word calculated based on a length of the word and a number of remaining characters up to an upper limit number of characters of a summary, the first score being output in response to input of input text to the model, calculating, based on a combined score generated by the combining, a distribution of generation probabilities of the word, and updating, based on the calculated distribution of the generation probabilities, parameters in the model.
Claims6
A computer-implemented learning method comprising:
- obtaining, for each of one or more words included in a dictionary of a model, a combined score by combining a first score and a second score, the first score being a score outputted from the model in response to inputting a character string to the model, the second score being a score calculated based on a length of that word and a number of remaining characters up to an upper limit number of characters of a summary;
- calculating, based on the combined score, a probability distribution of the one or more words;
- and
- updating, based on the calculated probability distribution, parameters in the model to generate a summary of the character strings,
- the second score for a first word having a shorter word length than the number of the remaining characters is higher than the second score for a second word having a longer word length than the number of the remaining characters, the first word and the second word being included in the one or more words.
A generating apparatus comprising:
- a memory;
- and
- a processor coupled to the memory and the processor configured to perform processing including:
- receiving a character string,
- obtaining, for each of one or more words included in a dictionary of a model, a combined score by combining a first score and a second score, the first score being a score outputted from the model in response to inputting a character string to the model, the second score being a score calculated based on a word length of that word and a number of remaining characters up to an upper limit number of characters of a summary,
- calculating, based on the combined score, a probability distribution of the one or more words, and
- updating, based on the calculated probability distribution, parameters in the model to generate a summary of the character strings,
- the second score for a first word having a shorter word length than the number of the remaining characters is higher than the second score for a second word having a longer word length than the number of the remaining characters, the first word and the second word being included in the one or more words.
A non-transitory computer-readable medium storing instructions which, when the instructions are executed by one or more computers, cause the one or more computers to perform processing, the processing comprising:
- receiving a character string;
- obtaining, for each of one or more words included in a dictionary of a model, a combined score by combining a first score and a second score, the first score being a score outputted from the model in response to inputting a character string to the model, the second score being a score calculated based on a word length of that word and a number of remaining characters up to an upper limit number of characters of a summary;
- calculating, based on the combined score, a probability distribution of the one or more words;
- and
- updating, based on the calculated probability distribution, parameters in the model to generate a summary of the character strings,
- the second score for a first word having a shorter word length than the number of the remaining characters is higher than the second score for a second word having a longer word length than the number of the remaining characters, the first word and the second word being included in the one or more words.
Dependencies are inferred from the claim text.