Method of deriving quality improvement requirements
The present disclosure relates to a method of deriving quality improvement requirements. The method according to some embodiments may include receiving a plurality of evaluation information for a first product from an evaluation information database, inputting the plurality of evaluation information into a key phrase extraction model and determining key phrases for the plurality of evaluation information based on an output of the key phrase extraction model, grouping some of the key phrases for the plurality of evaluation information into multiple groups and storing information on each of the multiple groups in a memory, ranking the key phrases included in each of the multiple groups and storing information associated with the ranks of the key phrases included in each of the multiple groups in the memory and determining top n key phrases from each of the multiple groups as quality improvement requirements.
1 . A method of deriving quality improvement requirements, performed by a computing system, the method comprising:
receiving a plurality of evaluation information for a first product from an evaluation information database;
inputting the plurality of evaluation information into a key phrase extraction model and determining key phrases for the plurality of evaluation information based on an output of the key phrase extraction model;
grouping some of the key phrases for the plurality of evaluation information into multiple groups and storing information on each of the multiple groups in a memory;
ranking the key phrases included in each of the multiple groups and storing information associated with the ranks of the key phrases included in each of the multiple groups in the memory; and
determining top n key phrases from each of the multiple groups as quality improvement requirements, based on ranking results for the key phrases included in each of the multiple groups.
2 . The method of claim 1 , further comprising:
crawling the plurality of evaluation information for the first product from the Internet.
3 . The method of claim 1 , wherein the key phrase extraction model includes a transformer model trained based on evaluation information for multiple products.
4 . The method of claim 1 , wherein the determining the key phrases for the plurality of evaluation information, comprises extracting n candidate phrases from each of the plurality of evaluation information, ranking the candidate phrases based on weighted averages and weighted sums of the candidate phrases and a number of times each of the candidate phrases is included in the plurality of evaluation information, and determining the key phrases for the plurality of evaluation information based on ranking results for the candidate phrases.
5 . The method of claim 1 , further comprising determining key words for the key phrases for the plurality of evaluation information, based on the key phrases for the plurality of evaluation information.
6 . The method of claim 5 , wherein the determining the key words for the key phrases for the plurality of evaluation information, comprises extracting nouns and roots contained in each of the key phrases for the plurality of evaluation information.
7 . The method of claim 5 , wherein the grouping some of the key phrases for the plurality of evaluation information, comprises extracting topic words for a plurality of topics from the plurality of evaluation information, and grouping some of the key phrases for the plurality of evaluation information into multiple groups based on similarities between the key words and the topic words.
8 . The method of claim 1 , wherein the ranking the key phrases included in each of the multiple groups, comprises ranking key phrases included in a first group, among the multiple groups, based on weighted averages and weighted sums of the corresponding key phrases and a number of times each of the corresponding key phrases is included in the first group.
9 . A computing system comprising:
at least one processor; and
a memory storing one or more instructions,
wherein by executing the stored instructions, the at least one processor performs operations of: receiving a plurality of evaluation information for a first product from an evaluation information database; inputting the plurality of evaluation information into a key phrase extraction model and determining key phrases for the plurality of evaluation information based on an output of the key phrase extraction model; grouping some of the key phrases for the plurality of evaluation information into multiple groups and storing information on each of the multiple groups in a memory; ranking the key phrases included in each of the multiple groups and storing information associated with the ranks of the key phrases included in each of the multiple groups in the memory; and determining top n key phrases from each of the multiple groups as quality improvement requirements, based on ranking results for the key phrases included in each of the multiple groups.
10 . The computing system of claim 9 , wherein the operation of determining the key phrases for the plurality of evaluation information, comprises operations of: extracting n candidate phrases from each of the plurality of evaluation information; ranking the candidate phrases based on weighted averages and weighted sums of the candidate phrases and a number of times each of the candidate phrases is included in the plurality of evaluation information; and determining the key phrases for the plurality of evaluation information based on ranking results for the candidate phrases.
11 . The computing system of claim 9 , wherein the at least one processor further performs an operation of determining key words for the key phrases for the plurality of evaluation information, based on the key phrases for the plurality of evaluation information.
12 . The computing system of claim 11 , wherein the operation of determining the key words for the key phrases for the plurality of evaluation information, comprises an operation of extracting nouns and roots contained in each of the key phrases for the plurality of evaluation information.
13 . The computing system of claim 11 , wherein the operation of grouping some of the key phrases for the plurality of evaluation information, comprises operations of: extracting topic words for a plurality of topics from the plurality of evaluation information; and grouping some of the key phrases for the plurality of evaluation information into multiple groups based on similarities between the key words and the topic words.
14 . The computing system of claim 9 , wherein the operation of ranking the key phrases included in each of the multiple groups, comprises an operation of ranking key phrases included in a first group, among the multiple groups, based on weighted averages and weighted sums of the corresponding key phrases and a number of times each of the corresponding key phrases is included in the first group.
15 . The computing system of claim 9 , wherein the at least one processor further performs an operation of crawling the plurality of evaluation information for the first product from the Internet.
16 . The computing system of claim 9 , wherein the key phrase extraction model includes a transformer model trained based on evaluation information for multiple products.