Big Data-driven In situ Design and Digital Preservation of Cultural Heritage in Jiangsu Rural Areas

Word Frequency-Inverse Document Frequency (TF-IDF) is used to measure the importance of a word in a document set or one of the documents in a corpus. The importance of a word is proportional to how often it appears in a document. A word is considered to be well typed if it occurs TF frequently in one document and rarely in others [18]. Therefore, the TF-IDF algorithm tends to keep the words that have the ability to classify and remove common words.

TF-IDF is TF multiplied by IDF. Where TF refers to the frequency of occurrence of a specified word in that corpus, normalized. Then the TF value is expressed as: (1) TFi,j=ni,jΣknk,j \[T{{F}_{i,j}}=\frac{{{n}_{i,j}}}{{{\Sigma }_{k}}{{n}_{k,j}}}\]

In Equation (1), the numerator is n_ij is the frequency of occurrence of word T_i in document D_j and the denominator is the frequency sum of all words.

Inverse Document Frequency measures the general importance of a word by the distribution of documents containing a word in the document set. If the fewer the documents containing the word, the larger the IDF, the better the categorization ability of the word. The IDF for a particular word is calculated as: (2) IDFi=lgiDI1+kj:Ti∈Dj}I \[ID{{F}_{i}}=\lg \frac{iDI}{1+kj:{{T}_{i}}\in {{D}_{j}}\}I}\] Where, |D| is the total number of documents in the corpus, 1+|{j:T_i,D_j}| is the number of documents containing the word finally TF-IDF is obtained from the product of TF and IDF: (3) TFIDFi,j=TFi,j×IDFi \[TFID{{F}_{i,j}}=T{{F}_{i,j}}\times ID{{F}_{i}}\]

LDA is a topic model that outputs the topic of each document in the document set as a probability distribution, which is used to infer the topic distribution of the document, and is often applied to the classification of text. The study establishes a topic model for Jiangsu countryside in terms of place-based design and cultural heritage protection through the LDA model. The training of the model can be described as such a dynamic chain: (4) P(Words|Text)=P(Words|Topic)P(Topic|Text) \[P(\text{Words}|\text{Text})=P(\text{Words}|\text{Topic})P(\text{Topic}|\text{Text})\]

The LDA model assumes that topics can be represented by a lexical distribution and an article can be represented by a topic distribution. The LDA topic model is a three-layer Bayesian network structure consisting of documents, topics, and words. When processing with python, only the number of topics, iterations, and Dirichlet parameters need to be determined during model training, and there is no need to label the comment text labels [19].

In this paper, based on the sentiment lexicon of KnowledgeNet Hownet, we customize the sentiment lexicon through the SO-PMI (Sentiment Orientation Point Mutual Information Algorithm) method, expand the sentiment lexicon, and improve the accuracy of the sentiment classification of the text of the design of the rural in situ nature and the protection of cultural heritage in Jiangsu Province.

The SO-PMI algorithm consists of two parts: SO-PMI and PMI. The core idea is to judge the relevance of a word by calculating the probability of the word to be judged appearing at the same time as the benchmark word. If a word has a higher probability of appearing together with a positive word, then the word is a positive sentiment word, i.e., in this paper, the sentiment tendency of a word is judged by judging the relevance of the word to positive or negative words [20].

Where PMI is calculated as: (5) PMI=log2(P(word1,word2)P(word1)P(word2)) \[PMI=lo{{g}_{2}}(\frac{P(word1,word2)}{P(word1)P(word2)})\]

If PMI>0, it indicates that the two terms are related, and the larger the value, the stronger the correlation. If PMI=0 or PMI<0 it means that the two terms are independent or unrelated.

The formula for SO-PMI is: (6) SO−PMI(word)=∑i=1num(pos)PMI(word,posi)−∑i=1num(neg)PMI(word,negi) \[SO-PMI(word)=\sum\limits_{i=1}^{num(pos)}{P}MI(word,po{{s}_{i}})-\sum\limits_{i=1}^{num(neg)}{P}MI(word,ne{{g}_{i}})\]

In Equation (6), num(pos) and num(neg) refer to the number of positive sentiment words and negative sentiment words of the benchmark dictionary, respectively. In this paper, the positive and negative emotions words from Hownet’s emotion lexicon are used as benchmark words respectively to calculate the SO-PMI values of the words in the text comments. If the SO-PMI with positive emotion words is >0, the word is a positive word. If the SO-PMI with negative sentiment words is <0, the word is judged as a negative sentiment word.

The emotional words that modify the symbolic elements of the image of the Jiangsu countryside in locality design and cultural heritage protection were extracted from the collected texts.

The emotion words corresponding to the symbols of the locality design texts are shown in Table 5. According to the corresponding emotion words, it can be judged that the current local design of Jiangsu countryside is not good enough in the aspects of rural transformation (A1), village name participation (A7), sustainable development (A8) and public space (A10).

The corresponding emotion words of cultural heritage protection text symbols are shown in Table 6. According to the corresponding emotion words, it can be judged that the current Jiangsu countryside needs to be protected and inherited in terms of intangible cultural heritage (B2), farming culture (B4), festival culture (B5), a hundred craftsmen and women (B7), and folk arts and culture (B8).

After extracting sentiment words, the next step will be to use the sentiment dictionary to calculate the SO-PMI value of each sentiment word. After splitting the text into words, the Python program is used to write the code to import the text data into the Hownet emotion dictionary to analyze the corpus, and based on the emotion classification of the semantic words in the training set, the emotion classification is mapped to each text utterance, and the SO-PMI values corresponding to the image symbols of Jiangsu countryside locality design and cultural heritage preservation under each category are calculated.

The SO-PMI values of the image symbols of the in-situ design of Jiangsu countryside are shown in Fig. 1. The total SO-PMI value of the 10 image categories is -85.4, with an overall negative sentiment, indicating that the public is not satisfied with the current in situ design of Jiangsu countryside. The image category with the lowest SO-PMI value is countryside renovation (-58.6), so Jiangsu countryside needs to carry out in situ design in Lixiang Village, Pan’an Lake Wetland Park, and the People’s Commune, People’s Commune and other places for in situ design.

Figure 1.

SO-PMI value of the image symbol of the rural area of Jiangsu

The SO-PMI values of the cultural heritage protection image symbols in rural Jiangsu are shown in Figure 2. The total SO-PMI value of the 10 image categories is -44.3. The image category with the lowest SO-PMI value is intangible cultural heritage (-61.8), and the ones with higher SO-PMI values are Baikong handicrafts (78.5) and folk arts and culture (55.9). Therefore, Jiangsu villages need to digitally protect intangible cultural heritage and culturally transmit the hundred crafts and folk arts.

Figure 2.

SO-PMI value of the image symbol of rural cultural heritage in Jiangsu

Combined with the characteristics of Jiangsu’s rural cultural heritage, there are five types of digital preservation methods: information database, digital archaeology, virtual interactive design, digital museum, and digital publication.

Digital technology has advantages in data collection and organization of Jiangsu rural cultural heritage, cloud storage and management can provide database technology support for the digital protection of Jiangsu rural cultural heritage, and Jiangsu rural cultural heritage related historical documents, local county records, folklore, interviews with relevant people, heritage images, mapping data, audio (video), animation and so on are the contents of the database, according to the data. According to the content of the data, the corresponding categorized information database can be established. For example, the database of folk cultural activities in the cultural heritage of rural villages in Jiangsu, in addition to storing relevant textual information, it can also adopt the way of video to produce documentaries of performances of liuqin opera and bangzi opera for living protection. To build a database of traditional handicrafts, the production process should be recorded, and process animation can also be added to supplement it, while photographs of finished products should be taken to improve the work library. In short, the database should be built with other media networks to achieve a comprehensive docking, adhere to the principle of sharing and openness, which will be conducive to the permanent storage and sharing of cultural heritage in rural Jiangsu, to facilitate the later updating and application of the popularization of rural culture in Jiangsu and inheritance of an important guarantee.

Digital technology can be digitized for all kinds of cultural relics archaeology and restoration, the use of digital technology to obtain three-dimensional data of cultural relics, reverse reverse the missing part of the damaged cultural relics of the data, three-dimensional modeling, and then the real cultural relics for effective restoration. For example, Jingshan Bridge, Guangyun Cang, Jiangsu Ancient City, etc., due to the Yellow River flooding attack, many architectural heritage was buried underground or severely damaged. This kind of relics is very suitable for the use of three-dimensional laser scanning technology on the relics of the relevant information collection, combined with the literature and historical data based on three-dimensional modeling in the software, and combined with virtual reality technology and mixed reality technology for its virtual reproduction, to restore the original appearance of the monument.

In the information database and digital archaeology on the basis of virtual interactive design, the use of digital photography, 3D scanning, engineering drawing 3D modeling and virtual reality technology and other means to create a virtual space scene. Such as the use of AnyCAD and other software to carry out an all-round observation of the three-dimensional model, but it can also be enlarged to observe the internal structure of the space scene and architectural details. Or design interactive games, so that the public can immersively experience the production process of Jiangsu incense buns and Jiangsu paper-cutting, in order to increase the sense of immersion, experience and interaction during online and offline tours, and to experience the charm of the heritage of rural Jiangsu in a three-dimensional and all-round way.

There is no physical museum in rural Jiangsu, and the establishment of a digital museum can save part of the expenditure on building a museum, and it can also have the effect of cultural promotion, i.e., the use of digital methods to project the tangible and intangible cultural heritage of rural Jiangsu from the physical material space to the virtual space. When establishing digital museums, virtual tour entrances can be established, and then combined with digital audio-visual, stereoscopic display, holographic projection, VR, AR, virtual roaming and other technical means to allow the public to tour the museum online on the computer side or cell phone side for virtual scene experience. Scattered on both sides of the Jiangsu Province, many cultural relics, few young people will go to the field to visit, but through the digital means to improve the cultural experience, show the cultural connotation, will let them because of the online cultural experience and the field of the offline visit interest, so as to enter into the depth of interaction with the culture to come.

Based on the preliminary data, the production of communication carriers such as e-books, animated videos, digital journals, APPs, websites, etc. for the promotion of cultural heritage in rural Jiangsu will be carried out. Utilizing the public’s fragmented time, we establish mobile Internet communication supported by QR code technology, use major information platforms to display digital reading materials converted into QR codes, and provide the public with instant access to illustrated explanations and animated videos of cultural heritage in a lively form.