Apparatus and method for searching interactive data through semantic index
AssigneeCoxwave
InventorSang yeop Kim
Life of this patent
Estimating the term…
Abstract
An embodiment may provide a method of searching interactive data through a semantic index, the method including acquiring interactive data consisting of at least one message; generating a semantic index by converting the message into a sentence structure having a single meaning, wherein the sentence structure is the same for all messages; acquiring a query for a user to search for interactive data; obtaining similarity between the semantic index and the query; returning interactive data corresponding to the query based on the similarity; and providing the user with the interactive data corresponding to the query.
Claims5
A method of searching interactive data through a semantic index, the method comprising:
- acquiring interactive data consisting of at least one message;
- generating a semantic index by converting the message into a sentence structure having a single meaning, wherein the sentence structure is the same for all messages;
- acquiring a query for a user to search for interactive data,
- wherein the acquiring of the query comprises:
- generating a query semantic index indicating a semantic index for the query;
- and
- converting the query semantic index into a query embedding vector;
- obtaining similarity between the semantic index and the query,
- wherein the obtaining comprises:
- acquiring a first embedding vector, which is converted from the interactive data, and obtaining a first similarity between the first embedding vector and the query embedding vector;
- acquiring a second embedding vector, which is converted from the message included in the interactive data, and obtaining a second similarity between the second embedding vector and the query embedding vector;
- acquiring a third embedding vector, which is converted from a message semantic index indicating a semantic index for the message, and obtaining the third similarity between the third embedding vector and the query embedding vector;
- and
- calculating scores through the first similarity, the second similarity, and the third similarity, by applying a first weight to the first similarity, a second weight to the second similarity, and a third weight to the third similarity;
- extracting the interactive data having the highest score among the calculated scores;
- and
- providing the user with the extracted interactive data.
An apparatus for searching interactive data through a semantic index, the apparatus comprising:
- an interactive data acquisition part configured to acquire interactive data consisting of at least one message;
- a semantic index generation part configured to generate a semantic index by converting the message into a sentence structure having a single meaning, wherein the sentence structure is the same for all messages;
- a query acquisition part configured to acquire a query for a user to search for interactive data, by performing:
- generating a query semantic index indicating a semantic index for the query;
- and
- converting the query semantic index into a query embedding vector;
- a similarity obtaining part configured to obtain similarity between the semantic index and the query by performing:
- acquiring a first embedding vector, which is converted from the interactive data, and obtaining a first similarity between the first embedding vector and the query embedding vector;
- acquiring a second embedding vector, which is converted from the message included in the interactive data, and obtaining a second similarity between the second embedding vector and the query embedding vector;
- acquiring a third embedding vector, which is converted from a message semantic index indicating a semantic index for the message, and obtaining the third similarity between the third embedding vector and the query embedding vector;
- and
- calculating scores through the first similarity, the second similarity, and the third similarity, by applying a first weight to the first similarity, a second weight to the second similarity, and a third weight to the third similarity;
- an interactive data provision part configured to extract the interactive data having the highest score among the calculated scores, and provide the user with the extracted interactive data.
Dependencies are inferred from the claim text.