守田愛梨，荒木徹也，池上翔馬，丘上美紗子，住正宏，上田玲子，相良泰行，2015，
「チェダーチーズの粘弾性試験法に基づいた食感評価モデルの開発」
Development of Texture Evaluation Model Based on Viscoelastic Testing Methods for Cheddar Cheese
⇒_pdf
⇒CiNii
『日本食品工学会誌』16: 185-200. （査読有）
抄録
チェダーチーズの粘弾性計測パラメータと官能評価スコアの相互関係解析に基づいて，高精度で官能評価スコアを予測する機器試験法の組み合わせを選択する方法およびPLS解析法とANNモデリングを併用して品質と嗜好度を評価するモデルを提唱した．
供試サンプルとしてチェダーチーズ10品を選定し，破断，クリープおよびテクスチャの3試験法により粘弾性パラメータを取得した．他方，食感属性用語10語を含む官能評価用語38語を選定し，それらを使用した分析型官能評価と消費者嗜好度調査を実施して官能評価スコアおよび嗜好度を得た．
供試サンプルの2バイト咀嚼による荷重経時変化曲線に基づいて食感と粘弾性計測特性との時系列的対応関係を解釈し，口腔内で感じる食感変化を粘弾性試験法の組み合わせによって良好に予測する方法を提示した．
この方法ではPLS回帰分析の予測精度の比較によって食感評価スコアと良好な対応を示す粘弾性試験法の組み合わせを特定した．
さらには，それぞれの試験法の組み合わせで得られた粘弾性パラメータを適用して食感評価スコアおよび嗜好度を予測するANNモデルを構築した．
この方法の導入によって，食感評価スコアの予測に最も寄与する粘弾性試験法の組み合わせを選択してANNモデルを操作することが可能となった．
Novel methodologies have been developed to predict the values of sensory evaluation score by selecting the combination of measurement methods for grouping viscoelastic parameters, and coupled PLS and ANN modeling have been presented as a quality evaluation model based on the relationship analysis between viscoelastic parameters and sensory scores for Cheddar cheese.
Ten Cheddar cheeses were selected as test samples and viscoelastic characteristics of the samples were obtained by three measurement methods of fracture, creep and texture tests.
Thirty-eight sensory terms including 10 terms of texture attribute were selected and then sensory evaluation using the terms as well as consumer palatability test were conducted for collecting both sensory and palatability scores, respectively.
The relationships between viscoelastic parameters and texture changes in human mouth were explainable by adapting the relationships between the viscoelastic characteristics and texture in the mastication process of the force vs. time curve obtained from the two bites process of the sample, and thus selection manner for the combinations of viscoelastic measurement methods corresponding to sensory texture scores were demonstrated by comparing the accuracy of PLS regression equations.
Furthermore, ANN models to predict sensory texture and palatability scores were successfully presented by applying each viscoelastic parameter obtained by the selected combinations of the measurement methods.
By the modeling process proposed in this study, ANN models were operational to utilize the optimal viscoelastic measurement methods contributing to the sensory and palatability scores.