Literature reading_ Research on the usefulness identification of tourism online comments based on semantic fusion of multimodal data (Chinese Literature)

Literature sources ： d , Li Gang , Chensijing , Mao Jin , Zhang Ji . Research on the usefulness identification of tourism online comments based on semantic fusion of multimodal data [J]. Journal of information ,2020,39(02):199-207.

Writing structure

introduction

Related work

Online comment usefulness identification

Multimodal data fusion

Online comment usefulness identification model based on multimodal data fusion

The model input layer is ： Given comment content with pictures and text
The model output layer is ： The value of the corresponding usefulness tag ,0 For useless comments , Greater than 0 For useful comments

Text semantic feature extraction

1、 Traditional word segmentation W2V

2、BERT
notes ：BERT Model word segmentation steps , The introduction of this document is too simple , When this technology is adopted in subsequent relevant literature , I'll complete it again .

Image semantic feature extraction

Picture label recognition

Label semantic vector extraction

Image and text feature fusion and model construction

Using text-based convolution neural network （TextCNN） And bidirectional long-term and short-term memory cycle neural network （bi-directional LSTM RNN）2 A deep learning model to build classifiers

Experiment and result analysis

experimental data

Descriptive statistics

Analysis of the number of comment pictures

Analysis of the number of useful votes

Usefulness classification experiment

Experimental process

Model implementation and result analysis

Results and discussion

---

problem ：
1、 How to define the usefulness and uselessness of comment content ？