Research Article
Recognition of Hotspot Words for Disease Symptoms Incorporating Contextual Weight and Co-Occurrence Degree
| Symbol | Description |
| d | Disease description | D | The set of disease descriptions | Em | Masked vector for d | Eo | Original vector for d | MeanSemSim (w, d) | The mean value of semantic similarity between w and words in d | TF-IDF (w, d, D) | The value of TF-IDF for the word w in d | α | The weight balance factor | si | The ith disease symptom word | G | Disease symptom word association graph | si↔sj. | si and sj satisfy co-occurrence association | Nco (si, sj) | The weight of the co-occurrence edge | vi(K) | Vector feature representation of the ith node in the Kth time step | P(K) | The predicted value at Kth time step | pi | The output result corresponding to the ith node | L(K) | The loss at Kth time step |
|
|