Meningkatnya perhatian global terhadap kendaraan listrik telah memicu diskursus publik yang masif di platform X, sehingga diperlukan metode analisis sentimen yang akurat untuk memahami persepsi masyarakat. Penelitian ini bertujuan mengevaluasi dan membandingkan kinerja empat model Deep Learning, yakni Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), dan model hibrida CNN-BiLSTM dengan mekanisme Attention. Metodologi penelitian melibatkan pengumpulan dari tweet berbahasa Indonesia periode Januari 2023 hingga Agustus 2025 yang dikumpulkan melalui web-scraping dan dilabeli menggunakan skor leksikon berbasis kuantil. Tahapan preprocessing meliputi case folding, noise removal, serta stemming menggunakan pustaka Sastrawi untuk menjamin kualitas data input. Hasil eksperimen menunjukkan bahwa model hibrida CNN-BiLSTM-Attention mencapai kinerja paling unggul dengan akurasi 75% dan macro F1-score 0,76. Model BiLSTM dan CNN masing-masing mencatat akurasi 73% dan 72%, sementara model LSTM standar mengalami kegagalan konvergensi (model collapse) karena ketidakmampuannya menangani kompleksitas data. Sinergi antara ekstraksi fitur lokal CNN dan pemahaman konteks sekuensial dua arah dari BiLSTM terbukti secara signifikan meningkatkan kemampuan klasifikasi, meskipun tantangan utama masih terletak pada ambiguitas antara kelas positif dan netral
Penelitian ini bertujuan untuk mengetahui dan menganalisis pengaruh Social Media Marketing, Service Quality, dan Electronic Word of Mouth terhadap Purchase Decision pada jasa fotografi dan videografi otomotif TPMedia.id di Kota Surabaya. Penelitian ini menggunakan pendekatan kuantitatif dengan metode explanatory research. Populasi penelitian berjumlah 217 pelanggan TPMedia.id dan sampel sebanyak 141 responden yang ditentukan menggunakan rumus Slovin. Teknik pengumpulan data dilakukan melalui penyebaran kuesioner dengan skala Likert. Analisis data menggunakan regresi linear berganda dengan bantuan program SPSS.Hasil penelitian menunjukkan bahwa Social Media Marketing berpengaruh positif dan signifikan terhadap Purchase Decision dengan nilai t hitung sebesar 3,952 dan signifikansi 0,000. Service Quality berpengaruh positif dan signifikan terhadap Purchase Decision dengan nilai t hitung sebesar 2,818 dan signifikansi 0,006. Electronic Word of Mouth juga berpengaruh positif dan signifikan terhadap Purchase Decision dengan nilai t hitung sebesar 2,298 dan signifikansi 0,023. Hasil uji koefisien determinasi menunjukkan nilai R Square sebesar 0,583, yang berarti bahwa Social Media Marketing, Service Quality, dan Electronic Word of Mouth mampu menjelaskan Purchase Decision sebesar 58,3%, sedangkan sisanya sebesar 41,7% dijelaskan oleh variabel lain di luar penelitian. Variabel yang paling dominan memengaruhi Purchase Decision adalah Social Media Marketing dengan nilai Standardized Beta sebesar 0,354.Berdasarkan hasil penelitian dapat disimpulkan bahwa Social Media Marketing, Service Quality, dan Electronic Word of Mouth secara parsial maupun simultan berpengaruh positif dan signifikan terhadap Purchase Decision pada jasa fotografi dan videografi otomotif TPMedia.id di Kota Surabaya.
This study evaluates the implementation and impact of equalizing administrative positions to functional positions (PermenPANRB No. 17/2021) within the Tuban Regency Government. Employing a qualitative descriptive-analytical approach, it integrates William N. Dunn’s evaluation framework, Agile Governance, and Performance Theory. The findings indicate that while the policy succeeded at the macro-administrative level by delayering hierarchies and streamlining budgets, it generated severe micro-operational issues. Efficiency was hampered by high adaptation costs, and a significant competence mismatch affected 26.50% of civil servants. Internal equity collapsed as equalized officials shouldered a double burden of functional and administrative tasks without adequate career paths. Furthermore, internal responsiveness was critically low due to culture shock, latent resistance, and career stagnation anxieties. The policy also proved inappropriate for general administrative support units. Collectively, these issues triggered post-power syndrome and role ambiguity, hindering organizational agility. To mitigate these disruptive impacts, the study recommends immediate competence calibration, precise demarcation of functional and administrative duties, and the expansion of career pathways through structured talent management at the local level.
Road pavement damage such as cracking and plastic deformation is influenced by the quality of binding materials and improper mixture composition, therefore an asphalt mixture innovation is required to improve pavement performance. The use of Cold Paving Hot Mix Asbuton (CPHMA) with Asbuton B30/50 is considered an alternative for utilizing local materials; however, improper asphalt content control may cause the mixture to become excessively rigid or overly plastic. Therefore, this study aims to analyze the effect of asphalt content variation on the Marshall characteristics of CPHMA mixtures using Asbuton B30/50 and to determine the Optimum Asphalt Content (OAC). The study was conducted using asphalt content variations of 6%, 6.5%, 7%, 7.5%, and 8%. Testing was carried out based on the 2018 Revised 2 Bina Marga General Specifications using the Marshall method to obtain stability, flow, Void in Mix (VIM), Void in Mineral Aggregate (VMA), Void Filled with Bitumen (VFB), and residual stability parameters. The results showed that increasing asphalt content caused the stability value to increase from 496 kg at 6% asphalt content to 556 kg at 7%, then decrease to 517 kg at 8%. The values of flow and VFB increased, while VIM and VMA decreased due to the reduction of air voids in the mixture. Increasing asphalt content made the mixture denser and improved aggregate bonding, but reduced the mixture porosity. Based on the Marshall analysis, 7% asphalt content was selected because it provided the best balance between stability, flexibility, and mixture durability.
Work-life balance is an essential aspect of the well-being of working mothers, particularly those with children in the golden age (0–5 years), as they face substantial demands from both work and childcare responsibilities. The ability to maintain a balance between work and family life is influenced by various factors, including social support and family-to-work enrichment. This study aimed to examine the relationship between social support and work-life balance, with family-to-work enrichment serving as a mediating variable among working mothers with one child in the golden age (0–5 years). This study employed a quantitative approach with a correlational research design. The participants consisted of 253 working mothers with one child aged 0–5 years, selected using a purposive sampling technique. The research instruments included the Social Support Scale, the Family-to-Work Enrichment Scale, and the Work-Life Balance Scale. Data were analyzed using Hayes' PROCESS Macro Model 4. The findings revealed that social support was positively associated with work-life balance. In addition, social support was positively associated with family-to-work enrichment, and family-to-work enrichment was positively associated with work-life balance. However, the mediation analysis indicated that family-to-work enrichment did not mediate the relationship between social support and work-life balance among working mothers with children in the golden age (0–5 years). These findings suggest that efforts to enhance work-life balance among working mothers should focus on strengthening social support from family members, spouses, coworkers, and organizations. Furthermore, future research is encouraged to investigate other mechanisms that may better explain the relationship between social support and work-life balance.
