ZTABS builds NLP systems for Barcelona enterprises — multilingual tourism text analytics, Spanish and Catalan legal document intelligence, and pharmaceutical NLP that extracts structured data from the multilingual documents across Catalonia's internationally connected economy.
Barcelona enterprises operate at the intersection of three languages — Catalan, Spanish, and English — creating unique NLP challenges that no other European city faces in quite the same way. Catalan (spoken by 10M+ people, official language of Catalonia alongside Spanish, with its own literary tradition and regulatory corpus) is a distinct Romance language, not a dialect of Spanish — it shares features with both Spanish and French but has its own morphology, syntax, and vocabulary. Barcelona business text regularly mixes all three languages: a hotel guest review might be in English, the staff response in Spanish, and the internal quality report in Catalan. A legal contract between Barcelona companies may use Catalan legal terminology (following the Codi Civil de Catalunya — Catalonia's own civil code, which differs from the Spanish Código Civil on significant points including property law and family law). A pharmaceutical clinical trial report from a Barcelona site uses Spanish medical terminology but references AEMPS (Spanish drug authority) guidelines in Castilian Spanish and EMA guidelines in English. This trilingual reality makes Barcelona a uniquely demanding NLP market. Spanish NLP itself is technically interesting: Spanish morphology (gendered nouns, complex verb conjugations with irregular forms, subjunctive mood in everyday business text) and Spanish business vocabulary create processing requirements distinct from English NLP. Catalan NLP is underserved — while Spanish has benefited from significant NLP investment (Spanish is the world's 4th most spoken language), Catalan language technology lags behind, creating both challenges and opportunity for Barcelona enterprises. The regulatory environment adds linguistic complexity: AEPD (data protection) publishes in Spanish, the Generalitat de Catalunya publishes regulations in Catalan, and EU regulations are referenced in English — a Barcelona enterprise must process regulatory text in all three languages.
Barcelona's NLP demand spans tourism and hospitality, legal services, pharmaceutical and biotech, and technology. Tourism text analytics: Barcelona's tourism industry (12M+ visitors annually from 100+ countries) generates enormous multilingual text volumes. Guest reviews: 2M+ reviews annually on Booking.com, TripAdvisor, Google, and other platforms — in 20+ languages, with Spanish, English, French, German, and Italian dominating. Review analysis requires multilingual sentiment detection, aspect extraction (cleanliness, location, service, value), and trend identification. Guest communications: hotels, restaurants, and attractions manage customer inquiries, complaints, and feedback in multiple languages. Tourism operators need NLP that understands tourist-specific language — informal, often multilingual within a single message, with cultural context that affects interpretation (a German tourist's "ganz okay" indicates lukewarm satisfaction, while a Spanish tourist's "no está mal" might indicate genuine approval). Social media monitoring: Barcelona's tourism brand is discussed across global social platforms in dozens of languages — NLP monitors sentiment, identifies emerging issues, and tracks destination perception. Spanish and Catalan legal NLP: Barcelona's legal market operates under a dual legal system — Spanish national law (Código Civil, Ley de Enjuiciamiento Civil) and Catalan civil law (Codi Civil de Catalunya — Book I through VI, covering persons, family, succession, property, and obligations). Contracts between Catalan parties may be governed by Catalan civil law, creating NLP requirements for both Spanish and Catalan legal terminology. Pharma NLP: Barcelona's pharma cluster manages clinical documentation in Spanish and English — adverse event reports from Spanish sites, AEMPS regulatory correspondence in Spanish, and EMA submissions in English. Clinical trial site documentation from Spanish hospitals uses Spanish medical terminology (informe de alta — discharge summary, historia clínica — clinical record, consentimiento informado — informed consent). Technology and startup: Barcelona's tech ecosystem (22@Barcelona innovation district) generates predominantly English technical documentation but Spanish and Catalan business documentation — creating bilingual processing needs.
Our Barcelona NLP development specialises in trilingual (Catalan, Spanish, English) text processing. Trilingual NLP architecture: we build unified models that process Catalan, Spanish, and English text within a shared semantic space. Catalan and Spanish are related but distinct — our models capture both the similarities (enabling knowledge transfer between languages) and the differences (ensuring correct processing of language-specific constructs). Catalan-specific processing: Catalan morphology (article contractions: "del" = "de" + "el", apostrophes: "l'empresa", ela geminada: "l·l" in words like "col·laboració"), Catalan vocabulary (distinct from both Spanish and French — "empresa" = empresa in Spanish but "societat" in more formal Catalan, "contracte" = contrato in Spanish), and Catalan legal terminology (Codi Civil de Catalunya uses specific terminology: "arres" for earnest money, "censal" for a type of charge on property unique to Catalan law). Spanish NLP: morphological processing (verb conjugation analysis — critical for legal text where "deberá" implies future obligation, "hubiera debido" implies past breach), Spanish legal entity recognition (SA, SL, SLU company types), and Spanish regulatory vocabulary (AEPD, CNMV, Banco de España terminology). Tourism NLP: multilingual sentiment analysis calibrated for tourism-specific language across 20+ source languages, aspect-based sentiment extraction (isolating opinions about specific hotel/restaurant/attraction attributes), and cross-cultural interpretation (understanding that sentiment expression varies by culture and language). Integration with Barcelona enterprise systems: Spanish government platforms (AEAT, Sede Electrónica), Generalitat de Catalunya systems, tourism management platforms (PMS, OTA APIs), and pharma platforms (pharmacovigilance databases, clinical trial management systems). GDPR/LOPDGDD compliance per AEPD guidance.
Tourism / Hospitality
A Barcelona hospitality group (28 properties — hotels, restaurants, and tourist attractions across Barcelona, Costa Brava, and the Balearic Islands, 2,400 employees, €165M annual revenue) received 480,000 guest reviews annually across 8 platforms (Booking.com, TripAdvisor, Google, Expedia, Airbnb, TheFork/ElTenedor, social media, and direct feedback). The reviews arrived in 22 languages — Spanish (28%), English (24%), French (14%), German (10%), Italian (8%), Catalan (5%), and 16 other languages (11%). The problems: (1) Review analysis was manual and partial. A team of 6 guest experience analysts read and categorised reviews — but could only process reviews in Spanish, English, and French fluently. Reviews in German, Italian, Russian, Chinese, Japanese, Korean, and other languages were either ignored (for smaller volume languages) or machine-translated (losing nuance and cultural context). The team processed approximately 60% of total review volume. Coverage gap: the 40% of reviews not processed included German tourists (the 4th largest source market) — service issues affecting German guests were systematically invisible. (2) Sentiment analysis was inconsistent. Analysts rated reviews as positive, neutral, or negative — but ratings varied between analysts and across languages. Cultural differences in expression were not accounted for: a Japanese review describing a stay as "まあまあ" (maa-maa — so-so/acceptable) was sometimes rated "neutral" and sometimes "negative" depending on the analyst. A British review describing a hotel as "not bad" was sometimes rated "positive" and sometimes "neutral." (3) Aspect extraction was absent. The group could not systematically identify which aspects of the guest experience drove satisfaction or dissatisfaction. Was the decline in Barcelona city hotel ratings driven by room quality, cleanliness, noise, breakfast, staff service, or location? Without aspect-level analysis, improvement investments were based on manager intuition rather than data. (4) Competitive intelligence was limited. The group could not systematically compare their guest review performance against competitors — even though competitor reviews were publicly available on the same platforms. (5) Response management was slow and inconsistent. Guest review response (expected by platforms and guests) was managed by property-level staff — response rates varied from 85% (best property) to 32% (worst). Response quality varied from thoughtful engagement to generic templates. Average response time: 5.2 days.
We built a multilingual guest review NLP platform over 10 weeks. (1) Multilingual Review Processing Engine: processing reviews in all 22 languages. The engine: ingesting reviews from all 8 platforms via APIs and web monitoring, detecting review language automatically (including mixed-language reviews — common in Barcelona where tourists mix Spanish/English or French/Spanish), processing reviews through language-specific sentiment and extraction models (not machine translation followed by English NLP — native processing in each language preserving cultural nuance), and normalising results into a language-independent structured format (same output schema regardless of input language). Language coverage: 100% of reviews processed (from 60%). (2) Culturally-Calibrated Sentiment Engine: understanding sentiment expression across cultures. The engine: detecting sentiment with cultural calibration (understanding that German tourists express satisfaction more reservedly than Spanish tourists — a 4/5 rating from a German tourist indicates stronger satisfaction than a 4/5 from a Brazilian tourist), identifying sentiment intensity (from mild dissatisfaction to extreme frustration — calibrated by language and culture), detecting irony and sarcasm (particularly important in English and Spanish reviews — "lovely to be woken at 3am by construction" is negative despite positive vocabulary), and processing emoji and informal language (tourism reviews use informal register with emoji, abbreviations, and internet slang across all languages). Sentiment accuracy: 92.3% across all languages (measured against human-annotated validation set). (3) Aspect-Based Sentiment Engine: extracting opinions about specific experience attributes. The engine extracted sentiment for 18 aspects across all languages: room quality (size, comfort, furnishing, maintenance), cleanliness (room, bathroom, common areas), noise (street noise, guest noise, construction), breakfast and F&B (quality, variety, value), staff and service (friendliness, responsiveness, professionalism, language ability), location (proximity to attractions, transport access, neighbourhood quality), value (price-performance ratio), check-in/check-out (speed, efficiency, flexibility), amenities (pool, spa, gym, Wi-Fi), and 9 additional aspects. Aspect extraction accuracy: 89.7% across all languages. The engine identified the primary aspect driving negative sentiment in each review — enabling targeted improvement investments. (4) Competitive Intelligence Engine: monitoring competitor review performance. The engine: monitoring public reviews for 120 competitor properties across the same platforms, applying the same sentiment and aspect extraction to competitor reviews, generating competitive position analysis (the group's performance vs. competitors on each aspect, by property and market segment), and identifying competitive advantages and vulnerabilities (aspects where the group significantly outperforms or underperforms competitors). (5) Response Intelligence Engine: accelerating and improving review response. The engine: generating response drafts in the review language (Spanish, English, French, German, Italian — with appropriate formality and cultural tone), personalising responses based on review content (acknowledging specific praise, addressing specific complaints with relevant details), prioritising reviews for response (negative reviews, influential reviewers, reviews mentioning specific staff members), and tracking response rates and response quality across properties. Response generation: drafts generated in 30 seconds per review. Response rate: increased to 92% across all properties (from 32-85% range). Response time: 1.2 days average (from 5.2 days).
Review coverage: 100% of reviews processed (from 60%). The 40% previously unanalysed included 23,000 German-language reviews — analysis revealed 3 service quality issues specifically affecting German guests (breakfast variety, room temperature control, and communication in German at reception) that were invisible to the Spanish/English/French-focused manual process. Addressing these issues improved German guest satisfaction by 0.4 points on Booking.com (from 8.1 to 8.5 — a significant improvement in the competitive Barcelona market). Aspect-based insights: the platform identified that noise was the primary driver of negative sentiment across Barcelona city centre hotels — specifically construction noise during summer months. The group implemented construction-adjacent room management (blocking rooms near active construction sites during affected periods) — reducing noise-related complaints by 62%. Competitive positioning: systematic competitive monitoring revealed that the group outperformed competitors on staff friendliness but underperformed on breakfast quality — leading to breakfast improvement investment that improved breakfast sentiment scores by 35% within 6 months. Response management: 92% response rate (from 32-85% range). Response time: 1.2 days (from 5.2 days). Platform algorithms reward responsive properties — average platform ranking improved 0.3 positions across properties. Overall guest satisfaction: Booking.com average score improved from 8.3 to 8.7 across the portfolio — driven by data-informed improvements from NLP insights. Revenue impact: improved ratings and competitive positioning contributed to a 6.2% RevPAR increase across the portfolio. Guest experience team: not reduced — but scope expanded from reactive review reading to proactive experience improvement, using NLP insights to prioritise investments. Annual value: €10.2M (estimated revenue impact from improved ratings and RevPAR increase), €420K (response management efficiency), €280K (competitive intelligence value — data-driven improvement investments) = €10.9M. Development cost: €185,000. Payback: under 7 days.
Beyond positive/negative — aspect-based sentiment analysis that tells you exactly what customers love or hate about specific features, with domain-specific calibration.
Custom NER models trained on your domain to extract people, organizations, products, dates, amounts, and domain-specific entities from any text.
Multi-label document classification into your custom taxonomy with confidence scores and automated routing based on classification results.
Extractive and abstractive summarization of long documents, meeting transcripts, research papers, and customer conversations — preserving key information.
Identify and extract relationships between entities in text — connecting people to organizations, products to features, or symptoms to diagnoses.
Cross-lingual models that work across 100+ languages, with specialized fine-tuning for your target languages and domains.
Barcelona's NLP market includes hospitality technology companies, Spanish AI companies, and international review management platforms. Hospitality technology: ReviewPro (Barcelona-based guest intelligence platform — strong in hospitality review analytics but limited multilingual NLP depth), TrustYou (guest feedback platform), and Revinate (guest data platform). These platforms provide review aggregation and basic sentiment analysis but limited aspect-level extraction, no cultural sentiment calibration, and limited processing of less-common languages. Spanish AI: various Spanish NLP companies — most focused on general Spanish language processing rather than domain-specific applications. Catalan NLP: limited commercial offerings — most Catalan NLP is academic (Universitat de Barcelona, Universitat Politècnica de Catalunya). International review platforms: Medallia and Qualtrics (enterprise experience management — comprehensive platforms but not hospitality-specialised and limited multilingual NLP). The gap: hospitality review platforms provide basic multilingual sentiment but not the culturally calibrated, aspect-level extraction across 22 languages that Barcelona's international tourism market requires. No vendor provides trilingual (Catalan, Spanish, English) NLP for legal and business documents. ZTABS fills this with production multilingual NLP built for Barcelona's linguistically complex market.
NLP development for Barcelona enterprises ranges by scope. Domain-specific NLP model (multilingual sentiment, Spanish legal extraction, 6-10 weeks): €50,000-120,000. Department-level NLP platform (tourism review intelligence, legal document NLP, pharma text analytics, 10-16 weeks): €120,000-280,000. Enterprise NLP infrastructure (trilingual, multi-domain, production-grade, 14-24 weeks): €280,000-500,000. Barcelona NLP engineering rates: €85-170/hour. Spanish consulting firms: €150-300/hour. ZTABS delivers at competitive rates with trilingual (Catalan, Spanish, English) NLP expertise.
Verified reviews from real client engagements — sourced from our public testimonial archive and Clutch profile.
My experience is throughout positive. Communication, service, the short response times and the flawless execution of a challenging topic was absolutely great. ZTABS is definitely my first choice again.
Christian Neff
Bank Software Advisory · Bank Software Advisory
Fintech
Fantastic Agency! I couldn't fault them even if I tried. They always go above and beyond to meet your expectations and always produces quality work. Thank you ZTABS.
Stephanie Kal
CEO · Beauty Finder Australia
Marketplace
It has been great working with ZTABS. They bounce off the ideas along the way. Amazing Experience.
Joel Rowe
CEO · Drill Quoter
Marketplace
We don't just contract. We ship and operate our own software. 23 products in production.
Common questions about NLP & text analytics for Barcelona businesses
Catalan is a distinct Romance language — not a dialect of Spanish. Our models process Catalan natively: understanding Catalan morphology (article contractions, ela geminada), Catalan vocabulary (which differs from both Spanish and French), and Catalan legal terminology (Codi Civil de Catalunya). The models distinguish between Catalan and Spanish text (which look similar but carry different meanings in some contexts) and process each with language-appropriate rules.
ZTABS is a remote-first NLP & text analytics agency serving Houston businesses — including sentiment analysis, named entity recognition, document classification. We work with Energy & Oil/Gas, Healthcare & Biotech, Aerospace & Defense companies in Houston, TX via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
NLP & Text Analytics in New York, NYZTABS is a remote-first NLP & text analytics agency serving New York businesses — including sentiment analysis, named entity recognition, document classification. We work with Finance & Fintech, Media & Advertising, Fashion & Retail companies in New York, NY via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
NLP & Text Analytics in Los Angeles, CAZTABS is a remote-first NLP & text analytics agency serving Los Angeles businesses — including sentiment analysis, named entity recognition, document classification. We work with Entertainment & Media, E-commerce & DTC Brands, Gaming & AR/VR companies in Los Angeles, CA via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Development in Barcelona, SpainZTABS is a remote-first web development agency serving Barcelona businesses — including full-stack development, progressive web apps, api development. We work with Mobile & Telecom, Travel Tech, HealthTech companies in Barcelona, Spain via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Design in Barcelona, SpainZTABS is a remote-first web design agency serving Barcelona businesses — including ui/ux design, responsive design, custom interfaces. We work with Mobile & Telecom, Travel Tech, HealthTech companies in Barcelona, Spain via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Development in Barcelona, SpainZTABS is a remote-first AI development agency serving Barcelona businesses — including llm integration & fine-tuning, ai agents & automation, rag & knowledge systems. We work with Mobile & Telecom, Travel Tech, HealthTech companies in Barcelona, Spain via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
NLP & Text AnalyticsLearn more about our NLP & text analytics services nationwide.
Partner with ZTABS for expert NLP & text analytics in Barcelona. Get a free consultation today.
