COMMUNICATION SKILL EVALUATION FEEDBACK DEVICE, COMMUNICATION SKILL EVALUATION FEEDBACK METHOD AND COMMUNICATION SKILL EVALUATION FEEDBACK PROGRAM

• Main Authors: ASHIKAGA ERIKA, YAMADA TOMOHIRO
• Format: Patent
• Language: English; Japanese
• Published: 01.09.2016
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

PROBLEM TO BE SOLVED: To effectively improve the communication skill of a person.SOLUTION: In an analysis device 1, a language information analysis part 111, a meta language information analysis part 112 and a non-language information analysis part 113 analyze an utterance language, voice characteristics and behavior characteristics from imaging data obtained by photographing an object person during communication, and an evaluation analysis part 114 clusters characteristic vectors based on the degree of the analyzed characteristics in each utterance, respectively calculates the center vector of each cluster about a plurality of utterances that are evaluated as positive by an evaluation terminal 3, and selects a center vector that is the most similar to a vector of characteristics about an utterance that undergoes negative evaluation. Then, the evaluation analysis part 114 respectively calculates difference of the degree of each characteristic of an utterance corresponding to the selected center vector and the degree of each characteristic of an utterance that is evaluated as negative, and extracts a characteristic having the largest difference. After that, an analysis result output part 123 outputs the extracted characteristic as an improvement element to the utterance in which the extracted characteristic is evaluated as negative to a browsing terminal 4 of the object person.SELECTED DRAWING: Figure 2 【課題】人のコミュニケーションスキルを効果的に改善すること。【解決手段】解析装置１において、言語情報解析部１１１とメタ言語情報解析部１１２と非言語情報解析部１１３が、コミュニケーション中の対象者を撮影した撮像データから発話言語と音声の特徴と行動の特徴とを解析し、評価解析部１１４が、解析した特徴の度合いに基づく特徴ベクトルを発言毎にクラスタリングし、評価端末３で肯定評価された複数の発言に関する各クラスタの中心ベクトルをそれぞれ算出し、否定評価された発言に関する特徴のベクトルに最も類似する中心ベクトルを選択する。そして、評価解析部１１４は、選択した中心ベクトルに対応する発言の各特徴の度合いと否定評価された発言の各特徴の度合いとの差をそれぞれ算出し、差が最も大きい特徴を抽出する。その後、解析結果出力部１２３が、抽出した特徴を否定評価された発言に対する改善要素として対象者の閲覧端末４に出力する。【選択図】図２