首先,定义算法中涉及的所有变量:
-
大五人格特质(Big Five Personality Traits): [ \begin{aligned} O & : \text{开放性(Openness)} \ C & : \text{责任感(Conscientiousness)} \ E & : \text{外向性(Extraversion)} \ A & : \text{宜人性(Agreeableness)} \ N & : \text{神经质(Neuroticism)} \end{aligned} ]
-
情感状态(Mood VAD Tensor): [ \begin{aligned} V & : \text{效价(Valence)} \ A & : \text{唤醒(Arousal)} \ D & : \text{支配(Dominance)} \end{aligned} ]
-
好感度(Affinity): [ \text{Aff} \in [0, 1] ]
-
话题关联度(Topic Relevance): [ T \in [0, 1] ]
为每个主要因素分配权重,反映其在决策中的重要性:
[ w_p, \ w_m, \ w_a, \ w_t ]
其中:
[ \begin{aligned} w_p & : \text{人格特质的权重} \ w_m & : \text{情感状态的权重} \ w_a & : \text{好感度的权重} \ w_t & : \text{话题关联度的权重} \end{aligned} ]
假设满足权重之和为1:
[ w_p + w_m + w_a + w_t = 1 ]
每个人格特质对话题激发的影响通过对应的权重表示:
[ p = \sum_{i=1}^{5} w_{p_i} \cdot T_i = w_{O} \cdot O + w_{C} \cdot C + w_{E} \cdot E + w_{A} \cdot A + w_{N} \cdot N ]
其中 ( w_{p_i} ) 为各人格特质的权重,满足:
[ \sum_{i=1}^{5} w_{p_i} = 1 ]
情感状态通过VAD三个维度及其对应权重计算:
[ m = \sum_{j=1}^{3} w_{m_j} \cdot \text{Mood}_j = w_V \cdot V + w_A \cdot A + w_D \cdot D ]
其中 ( w_{m_j} ) 为VAD三个维度的权重,满足:
[ \sum_{j=1}^{3} w_{m_j} = 1 ]
将各部分得分按其权重相加,得到一个综合得分:
[ S = w_p \cdot p + w_m \cdot m + w_a \cdot \text{Aff} + w_t \cdot T ]
设定一个阈值 (\theta),用于判断是否激发话题:
[ \text{激发话题} = \begin{cases} \text{是}, & \text{如果 } S \geq \theta \ \text{否}, & \text{否则} \end{cases} ]
将上述步骤综合起来,完整的判断是否激发话题的数学公式如下:
[ \begin{aligned} p &= w_{O} \cdot O + w_{C} \cdot C + w_{E} \cdot E + w_{A} \cdot A + w_{N} \cdot N \ m &= w_V \cdot V + w_A \cdot A + w_D \cdot D \ S &= w_p \cdot p + w_m \cdot m + w_a \cdot \text{Aff} + w_t \cdot T \ \text{激发话题} &= \begin{cases} \text{是}, & \text{如果 } S \geq \theta \ \text{否}, & \text{否则} \end{cases} \end{aligned} ]
假设权重如下:
[ w_p = 0.3, \quad w_m = 0.2, \quad w_a = 0.3, \quad w_t = 0.2 ]
且各人格特质及VAD维度的权重为:
[ \begin{aligned} w_{O} &= 0.9, \quad w_{C} = 0.6, \quad w_{E} = 0.7, \quad w_{A} = 0.8, \quad w_{N} = 0.4 \ w_V &= 0.7, \quad w_A = 0.3, \quad w_D = 0.5 \end{aligned} ]
给定输入数据:
[ \begin{aligned} O &= 0.7, \quad C = 0.6, \quad E = 0.5, \quad A = 0.8, \quad N = 0.3 \ V &= 0.8, \quad A = 0.5, \quad D = 0.6 \ \text{Aff} &= 0.75, \quad T = 0.85 \ \theta &= 0.6 \end{aligned} ]
计算过程如下:
[ \begin{aligned} p &= 0.9 \times 0.7 + 0.6 \times 0.6 + 0.7 \times 0.5 + 0.8 \times 0.8 + 0.4 \times 0.3 \ &= 0.63 + 0.36 + 0.35 + 0.64 + 0.12 \ &= 2.10 \ &\Rightarrow \text{标准化}: \quad p = \frac{2.10 - \text{min}(p)}{\text{max}(p) - \text{min}(p)} = 0.75 \ \text{(例如)} \ m &= 0.7 \times 0.8 + 0.3 \times 0.5 + 0.5 \times 0.6 \ &= 0.56 + 0.15 + 0.30 \ &= 1.01 \ &\Rightarrow \text{标准化}: \quad m = \frac{1.01 - \text{min}(m)}{\text{max}(m) - \text{min}(m)} = 0.67 \ \text{(例如)} \ S &= 0.3 \times 0.75 + 0.2 \times 0.67 + 0.3 \times 0.75 + 0.2 \times 0.85 \ &= 0.225 + 0.134 + 0.225 + 0.170 \ &= 0.754 \ \end{aligned} ]
判断:
[ 0.754 \geq 0.6 \quad \Rightarrow \quad \text{激发话题: 是} ]