Persaingan antar perusahaan semakin sengit seiring waktu. Banyak perusahaan optimis akan performanya di masa depan, namun banyak juga perusahaan yang tidak yakin dapat bersaing. Kesulitan ini terutama dihadapi oleh perusahaan-perusahaan pada sektor dengan potensi besar yang diperebutkan seperti sektor video game. Perusahaan-perusahaan tersebut perlu memerhatikan persepsi / sentimen pelanggan agar dapat meningkatkan dan mempertahankan daya saingnya dalam jangka panjang. Persepsi ini umumnya ditunjukkan pelanggan melalui ulasan mereka terhadap produk perusahaan. Dengan demikian, perusahaan video game dapat mengidentifikasi kesempatan pengembangan atau peningkatan daya saing dengan memerhatikan persepsi pelanggan dari ulasan video game. Pembuatan model dengan metode Aspect Category Sentiment Analysis, salah satu bagian dari rangkaian metode Aspect-based Sentiment Analysis, dapat menjadi salah satu solusi agar perusahaan video game dapat melakukan hal tersebut. Oleh karena itu, penelitian ini memakai metode Aspect Category Sentiment Analysis secara unsupervised untuk membuat model sebagai solusi terhadap permasalahan perusahaan video game dan perusahaan lain yang kesulitan bersaing. Model tersebut dibuat dengan memanfaatkan vektor yang dihasilkan oleh model Word Embedding untuk merepresentasikan hubungan sentimen antar kata yang ada di ulasan video game. Hasil evaluasi menunjukkan bahwa model yang dibuat dapat merepresentasikan hubungan sentimen terhadap aspek video game yang diulas oleh pelanggan. Informasi ini kemudian dapat dipetakan agar perusahaan video game dapat mengidentifikasi kesempatan pengembangan atau peningkatan daya saing.

---

Competition among firms is intensifying over time. Many are optimistic about its future growth, but there are also many who are unsure about their own competitive capabilities. This pessimistic outlook is shared by a lot of firms in business sectors with many yet heavily contested business opportunities such as the video game sector. These firms must pay closer attention to the perception or sentiment of their customers so they can increase and maintain their long-term competitiveness. Such perception is generally expressed by customers through their product reviews. Hence, video game companies can identify product development opportunities or unknown competitive advantages/disadvantages by closely monitoring customer perception from video game reviews. Models created through Aspect Category Sentiment Analysis, a sub-discipline of Aspect-based Sentiment Analysis, can be a solution for video game companies to do such an endeavor. Therefore, this research created an unsupervised Aspect Category Sentiment Analysis model as a solution for video game companies and companies that face a similar problem. The model is created by utilizing the capability of word vectors from word embeddings to represent semantic relationships such as sentiment toward video game aspects that are mentioned in customer reviews. Thorough numerical and qualitative evaluation shows that the model can reliably represent such sentiment. Video game companies can then map the sentiment that is identified by the model to identify product development opportunities or unknown competitive advantages/disadvantages.