Recent comments and market analysis of luwei, soy sauce braise (近期滷味食品的評論及市場分析)

Through the self-made automated data capture system with specific keyword collections, we selected 242,528 articles retrieved in April 2018, and screened 704 reports or articles that touched on luwei, soy sauce braised foods. In these articles, we further extracted the comments directly related to luwei, soy sauce braised foods, and then analyzed the emotional tendencies of these comments using a "machine learning" trained "natural language processing, NLP" algorithm.
透過自製的自動化資料擷取系統搭配特定關鍵詞彙,我們將2018年4月擷取到的 24萬 2528篇文章中,篩選出 704篇與觸及滷味食品的報導或評論文章。在這些文章中,我們進一步萃取與滷味食品直接相關的評論,而後以經過「機器學習」訓練的「自然語言處理」的資料演算法,分析這些評論的情緒傾向。

In addition, the geographical region of Taiwan to which the comment belongs is also judged as another dimension of data analysis, trying to understand the similarities or differences of the comments made by Northern-Taiwan, Central-Taiwan and Southern-Taiwan consumers. In the quantitative analysis of emotional tendency, if we divide extreme negative emotions by score 0 and extreme positive emotions by score 100, we can obtain the following data results.
此外亦將評論所屬的台灣地理區域判斷出來,做為資料分析上的另一個維度,嘗試了解北中南消費者對於這項產品的評論異同。在情緒傾向的量化分析中,若以 0分為極端負面情緒、100分為極端正面情緒,我們可以獲得以下數據結果。

  • Emotional quantification and standard deviation for the Northern-Taiwan Reviews: 70.31 ± 30.29
    北台灣評論的情緒量化平均值與標準差:70.31 ± 30.29
  • Emotional quantification and standard deviation for the Central-Taiwan Reviews: 72.93 ± 31.25
    中台灣評論的情緒量化平均值與標準差:72.93 ± 31.25
  • Emotional quantification and standard deviation for the Southern-Taiwan Reviews: 70.71 ± 30.01
    南台灣評論的情緒量化平均值與標準差:70.71 ± 30.01


Under such a value, comparing the values ​​of any two geographical regions does not have significant differences. Similar scores also represent similar sentiments of users' comments on luwei, soy sauce braised foods in different regions during this period. However, when we analyzed the "word frequency" using the TF-IDF algorithm, we saw different discussion in between three geographic regions.
在這樣的數值之下,將任兩個地理區域的數值相比較,都是不具備顯著差異的。相近的分數也代表這段時間內、不同地區網友對滷味食品的評論情緒,大致上是相似的。然而當我們把評論文字以 TF-IDF演算法進行「詞頻」分析之後,卻看到三個地理區域有截然不同的討論內容。

In the Northern-Taiwan review above, consumers mainly focus on the discussion of ingredients and tastes. Although the "key opinion leaders, KOL" who volunteered to comment will not necessarily represent everyone's preferences, they can guide the audience and word of mouth. This result implies that the emphasis may be on ingredients and tastes, it will be more likely to cause Northern-Taiwanese consumers to discuss luwei, soy sauce braised foods.
在上圖的北台灣評論中,消費者主要著重於食材與口味的討論。雖然主動發表評論的「關鍵意見領袖」並不一定能代表所有人的偏好,但卻能引導觀眾與口碑的走向。此結果暗示倘若以食材與口味為重點,較能引起北台灣消費者對滷味食品的討論。

In the above review and analysis results, it can be seen that the taste of the sauce has become the comment focus of consumers in Central-Taiwan, and the ingredients may not be the key.
上圖的評論分析結果中,可以看出醬料的味道成了中台灣消費者的評論重心,而食材則可能不是關鍵。

In Southern-Taiwan consumer review, spicy taste and ingredients have become the focus of discussions in April this year.
在南臺灣的消費者評論中,麻辣口味與食材則成了今年四月的討論重心。

With such an analysis method, sufficient data support and continuous analysis, we have the opportunity to portray Taiwan consumers' preferences for a specific food. Then use the results of this analysis as a reference for decision making to assist the supply and promotion of products. Perhaps on the basis of such scientific analysis, it will have more opportunity for suppliers and consumers to achieve a common satisfaction. Welcome to contact us, if you have any word of mouth or public opinion analysis needs about any products.
藉由這樣的分析方法,足夠的數據資料支持與連續時間的分析之下,我們就有機會能描繪出台灣消費者對於特定飲食的喜好。進而以這樣的分析結果為決策參考,輔助產品的供應與推廣。或許在這樣的科學分析基礎下,更有機會使供應商和消費者取得共同滿意的交集點。如果您有任何產品的口碑或輿情分析需求,歡迎與我們聯繫。