<![CDATA[The stemming process in text preprocessing can ruin the meaning of words in Bahasa Indonesia text mining, potentially influencing the interpretation outcomes or the accuracy of machine learning models when processing complaint texts. Many Indonesians deliver their complaints by text, making this an important issue. Therefore, this research proposes the Be Raw Language (BRL) method for handling complaint texts. BRL generally circumvents several words that, when subjected to stemming, undergo changes in meaning. To ascertain whether a word undergoes changes and as a basis for analysis, this study employs a sentiment analysis approach utilizing 6,205 complaint text data sourced from community reviews concerning tourist destinations. Initially, these reviews are labeled as ground truth, and sentiment calculations are conducted. In the preliminary stage, the research findings indicate an accuracy rate of 60.23%. Subsequently, this study conducts an in-depth analysis on how words in the Indonesian language may change in meaning with the addition of prefixes or suffixes. Consequently, the concept of the BRL method emerges to analyze words without employing stemming and to delineate its approach in interpreting words along with their meanings. The study establishes three main rules for interpreting the meaning of a word or even phrases in Bahasa Indonesia texts to enhance accuracy. As a result, employing the BRL method increases the accuracy rate by 17.57% to 77.80%.]]>
