PRE2024 3 Group16

1. Project Overview

1.1 Group members

Name	Studentnumbers
Tim Hobbes	1798251
Eline Smit	1903071
Veselin Todorov	1789198
Georgi Kalchev	1855441
Ania Bărbulescu	1823612

1.2 Primary Objective

The primary objective of this project is to develop an AI-powered app designed to assist tourists in cities by providing real-time, interactive, and personalized support. The robot will serve as a multilingual city guide, offering tailored recommendations for local restaurants, helping tourists discover the best dining experiences in the area.

Our key objectives are to:

Enhance the tourist dining experience: Provide immediate, accurate, and personalized restaurant recommendations based on tourists' preferences, dietary restrictions, and the local food scene.
Promote local restaurants: Introduce tourists to local eateries, supporting small businesses and showcasing the diversity of regional cuisine.
Establish an effective means of communication: The social robot should ensure clear and effective communication with tourists, allowing them to easily find and understand restaurant options that best suit their needs.

2. Target Users and Stakeholders

2.1 Primary Users

The primary target group for this AI-powered app consists of domestic and international tourists exploring cities. In 2025, there are expected to be 1.4 billion international tourists (UN, 2024), with international tourist arrivals worldwide projected to grow by 3% to 5% in 2025 (UN, 2024). Our target group includes every tourist who does not speak the language of the city they travel to. Tourists often arrive in unfamiliar environments with limited knowledge of the city's layout, attractions, and cultural norms. Language barriers and the overwhelming amount of information available online can create a sense of disorientation and frustration. The app is designed to address these challenges by serving as an intelligent, interactive guide that enhances the tourist experience from the moment they arrive until their departure.

2.2 Secondary Users

A significant secondary target group, and the group poised to benefit most from a financial perspective, are local business owners and the city tourism board in the areas where the social robot operates. By enhancing the overall experience for tourists, these stakeholders can directly influence visitor satisfaction, which, in turn, contributes to long-term economic growth within the city.

The app, through its personalized recommendations and real-time information delivery, can promote local restaurants. By providing tailored suggestions based on user preferences, the app acts as a dynamic marketing tool, directing tourists toward hidden gems, seasonal events, and locally-owned establishments that might otherwise go unnoticed.

3. User study

To ensure that the app meets the needs of the tourists it's designed to assist, we conducted a small user study. (N=27)

The wiki experienced issues displaying images. You can find the results here: https://drive.google.com/file/d/1jHhfxTPQImgIav6mnnbPl-XDslybWFmp/view?usp=sharing

A more thorough summary can be found here: File:Summary Questionnaire.pdf

3.1 How Travelers Search for Information

The study revealed that most users rely on Google Search, TripAdvisor, and YouTube when planning their trips. Some also turn to AI tools like ChatGPT, particularly for gathering initial ideas and general research. However, users emphasized that while AI is useful for broad overviews, it is unreliable for specifics such as hotels, restaurants, and real-time travel details. The concern stems from outdated information and hallucinated (made-up) recommendations, which make AI-generated content less trustworthy for detailed planning.

Given this, our app will use AI primarily for overviews rather than for generating specific recommendations. To ensure accuracy, we will integrate the Google Search API to provide up-to-date food-related results, while AI will structure and refine this information for better readability.

3.2 Challenges Due to Language Barriers

One of the biggest difficulties travellers face is the inability to search in the local language of their destination. This limitation often restricts the accuracy and scope of the information they can find. Several destinations were identified as particularly difficult in terms of information accessibility, including Malaysia, Eastern Europe (Serbia, Romania, etc.), France, Japan, and Georgia. In places like France, much of the relevant travel content is only available in French, and in Japan, users reported challenges in understanding directions and finding reliable sources in English. In many cases, travellers either relied on Google Translate, luck, or asking locals to navigate these challenges.

3.3 Low Demand for Certain Travel Assistance Features

The study also indicated that transportation, currency exchange, event discovery, and budgeting are not major concerns for most travellers. Users generally do not struggle with finding public transport options, as they rely on Google Maps, local apps, or transportation websites. Similarly, budgeting tools were not seen as necessary, as travellers either manage expenses on their own or prefer not to be restricted by automated budgeting tools. Most participants also expressed a preference for spontaneity over structured, AI-generated trip plans or itineraries, indicating that automated planning features would not align with their travel habits.

Given this, our app will not include features related to public transportation, budgeting, event discovery, or planning as users prefer handling these aspects independently.

3.4 Finding Local Food: A Key Challenge

One of the most valuable insights from the study is that travellers actively seek summaries of local cuisine when visiting a new destination. However, finding authentic local food spots is not always straightforward. Users currently rely on a mix of methods, including Google and social media platforms (Instagram, YouTube, and Pinterest), recommendations from locals (friends, accommodation staff), and reading restaurant reviews in the local language. Many travellers also avoid tourist-heavy restaurants by looking for menus in the native language, reviews written by locals, and restaurants with mostly local customers.

Given these findings, our app will focus on food-related travel guidance by combining Google Search API with AI-generated summaries. This will allow users to access structured lists of local restaurants, information about traditional dishes, and food recommendations that overcome language barriers.

3.5 Interest in Hidden Gems and Less Touristy Spots

Many travellers expressed a strong interest in discovering hidden gems, including local restaurants, scenic viewpoints, lesser-known cultural sites, and unique events. To find these places, users often turn to Reddit (e.g., r/[destination]), Atlas Obscura, Google Maps, or local recommendations. Some travellers rely on exploring a city on foot, while others prefer reading travel blogs and niche forums.

While the app will initially focus on local food recommendations and cuisine summaries, it will be designed with flexibility in mind, allowing for future expansion based on demand. By leveraging the same methods used to identify authentic dining experiences, the app could be extended to help users discover local hidden gems such as scenic viewpoints, cultural sites, and unique attractions. This approach ensures that, if needed, the platform can seamlessly incorporate broader travel insights while maintaining its core functionality and reliability.

4. Market Research

Before designing our own app, we first searched for the AI Trip planners that are already on the market. To test their effectiveness, we used the same prompt across multiple AI trip planners: a 3-day trip to Amsterdam (March 7-9) with friends, focused on art and museums. The results showed that while these platforms are useful for general planning, they often lack localized insights, real-time event integration, and travel logistics considerations. For the images of the apps, follow this link: https://docs.google.com/document/d/10K9Qo3mwOFIJjQCA1PH3obCA7fBsB9op1oJJBdDsyHQ/edit?usp=sharing

4.1 Layla AI Trip Planner

Layla offers both free and paid versions. The free version gives a brief itinerary featuring must-see attractions, an estimated cost, and a hotel suggestion. While easy to use, its recommendations remain broad and lack hidden cultural gems or unique local experiences.

4.2 AI Trip Planner (BuildAI.space)

This tool provides a detailed itinerary with major attractions, dining spots, and a structured time plan. However, it fails to consider travel times, costs, and local recommendations, making it difficult for travelers to efficiently explore the city.

4.3 Mindtrip AI

Mindtrip offers a day-by-day plan with a built-in map, helping users visualize their trip. However, the recommendations focus only on the most popular tourist spots, with no emphasis on local or niche attractions.

4.4 Wonderplan

Wonderplan provides a general overview of Amsterdam and a list of the 26 most popular hotels, but lacks an actual structured itinerary. This makes it less useful for travelers seeking a planned cultural experience.

4.5 Tripadvisor AI Planner

Tripadvisor offers one of the most advanced AI trip planners, allowing users to customize their journey based on preferences, such as pet-friendly travel. However, it fails to account for travel distances between attractions, making its recommendations impractical for efficient city exploration.

4.6 Plantrip.io

A Dutch-based planner, Plantrip.io structures trips well and provides time-optimized plans. However, its suggestions remain broad, without offering local insights or real-time updates on cultural events.

5. Problem Statement

Tourism plays a pivotal role in the economies of many cities, drawing millions of visitors annually. However, navigating unfamiliar urban landscapes can be challenging for tourists. Common difficulties include overcoming language barriers, identifying authentic attractions, and accessing real-time, reliable information about transportation, local events, and cultural landmarks. Traditional tools—such as printed guidebooks, static information kiosks, and general-purpose mobile applications—often fall short in providing personalized, interactive, and immediate support. While human guides can offer valuable assistance, they are not always readily available or affordable, especially for solo travelers or those exploring lesser-known destinations. As global tourism continues to grow, the demand for innovative, tech-driven solutions becomes increasingly urgent.

5.1 Language Bias in LLMs

Search engines and AI-driven information retrieval systems are often perceived as neutral tools for accessing knowledge. However, a growing body of research suggests otherwise, revealing significant biases that favor dominant languages—particularly English—and culturally influential communities. This linguistic bias has profound implications, particularly for travelers seeking authentic local experiences. The difficulty in uncovering lesser-known, locally cherished destinations stems from search engines prioritizing globally dominant narratives over regionally significant information.

Luo, Puett, and Smith (2024), in their study A 'Perspectival' Mirror of the Elephant, investigate how search engines and AI-driven platforms such as Google, ChatGPT, YouTube, and Wikipedia reflect linguistic and cultural biases. They find that English-language searches overwhelmingly return content shaped by dominant Western perspectives, whereas searches conducted in local languages yield more culturally embedded information. Their research highlights how search algorithms reinforce a globalized knowledge hierarchy, making it harder for users—such as travelers seeking authentic local experiences—to discover region-specific insights. This supports the idea that tourists relying on English-language searches are more likely to be directed to mainstream, commercialized venues rather than authentic local spots.

Similarly, Rovira, Codina, and Lopezosa (2021), in their study Language Bias in the Google Scholar Ranking Algorithm, provide evidence of the systemic preference for English-language academic content. Their research reveals that Google Scholar ranks English-language papers significantly higher than those written in other languages, making non-English research nearly invisible in search results. This has broader implications beyond academia, as it reflects a general pattern in which English content is prioritized across various digital platforms. For travelers, this means that searches conducted in English will primarily surface popular, widely reviewed locations rather than smaller, locally favored spots that are discussed primarily in the local language.

The broader issue of search engine bias is extensively explored by Segev (2010) in Google and the Digital Divide: The Bias of Online Knowledge. Segev argues that Google systematically amplifies the knowledge and perspectives of dominant linguistic communities, creating a "digital divide" in which non-English content is suppressed. This digital divide has real-world consequences for those seeking culturally specific or community-driven insights, as it means that valuable local knowledge—such as recommendations for authentic dining experiences or hidden historical sites—remains largely inaccessible unless users explicitly search in the local language.

The problem extends beyond search engines to AI-powered tools, as highlighted by Sharma, Murray, and Xiao (2024) in their study Faux Polyglot: A Study on Information Disparity in Multilingual Large Language Models. Their research finds that large language models (LLMs) such as ChatGPT exhibit significant biases towards high-resource languages like English. When queried in a low-resource language, these models often provide translations or summaries based on English-language sources rather than drawing directly from native-language content. This creates a misleading sense of linguistic inclusivity while reinforcing the dominance of English-language information. For travelers relying on AI-driven recommendations, this means that even when searching in a local language, the results may still reflect globally dominant perspectives rather than authentic local insights.

The combined findings of these studies illustrate a crucial limitation of mainstream search engines and AI-driven information retrieval systems: their structural preference for English over other languages. This bias disproportionately affects travelers seeking authentic experiences, as it obscures regionally specific knowledge in favor of more globally recognized content. Addressing this issue requires a shift in search strategies—such as conducting searches in the local language—to uncover the true essence of a destination as experienced by its residents.

5.2 Language and Food

Language plays a crucial role in shaping cultural identity, and its influence extends deeply into local food traditions. The linguistic diversity found in various regions across the world reflects a society’s history, traditions, and culinary heritage. The relationship between language and food is evident in how local dishes are named, described, and preserved through generations. Dialects, in particular, play a significant role in maintaining the authenticity and uniqueness of regional cuisines.

5.2.1 The Role of Language in Culinary Traditions

Language serves as a bridge between food and culture, helping communities articulate their culinary practices and pass them down through generations. Research has shown that incorporating local dishes into language education can strengthen cultural roots and facilitate the dissemination of traditional knowledge (Tyas, 2017). For example, in Indonesia, adapting locally recognized culinary terms in learning materials helps introduce traditional dishes more effectively (Ainun et al., 2024). Similarly, ecolinguistic studies emphasize how language embodies cultural identity, as seen in the lexicons related to Balinese flora-based food processing (Umiyati & Pratama, 2021). The integration of food-related vocabulary in language learning highlights the deep connection between language and local cuisine, ensuring the survival of culinary heritage.

5.2.2 The Impact of Dialects on Local Food Terminology

Dialects contribute significantly to the linguistic representation of local food, influencing how dishes are named and described within specific communities. Studies on Vologda dialects in Russia illustrate how food traditions are embedded in speech genres such as dining conversations, memoirs, and culinary recipes (Zhandarova & Kryuchkova, 2023). Similarly, research on Sundanese dialects in Indonesia reveals lexical variations in food naming, encompassing changes in sounds, word formation, and semantics (Durahman & Badriah, 2022). These studies demonstrate that dialects not only reflect regional linguistic diversity but also serve as repositories of traditional culinary knowledge.

One of our team members, who is from Friesland, also discovered notable differences when searching for food-related information in the national language versus the local dialect spoken in Friesland. Some restaurants had distinct names in the dialect. This highlights how dialects influence access to and awareness of regional cuisine, reinforcing the importance of preserving linguistic diversity in culinary contexts. This example can be found by "Examples local languages".

In many cases, the specific dialectal terms for food ingredients and preparation methods are unique to a region and may not have direct translations in standard languages. This uniqueness underscores the importance of preserving dialects to maintain the integrity of local food traditions. However, the decline of dialect usage in some communities poses a challenge, as younger generations may lose access to the rich culinary lexicons of their ancestors (Zurriyati & Sinar, 2018). This shift threatens the continuity of traditional food knowledge, emphasizing the need for conscious efforts to document and promote dialectal culinary expressions.

5.2.3 The Cultural and Economic Implications

The interplay between language and local food extends beyond cultural identity to economic and tourism aspects. Many regions capitalize on their unique food culture to attract visitors, and the linguistic representation of local cuisine plays a key role in marketing traditional dishes. For instance, the use of indigenous terms for dishes in menus and food festivals enhances authenticity and fosters a deeper appreciation of regional heritage. In multilingual societies such as Cameroon and Indonesia, the coexistence of multiple languages and dialects allows for rich culinary exchanges while preserving local identities (Awah, 2021; Muth’im & Sutiono, 2024).

Moreover, linguistic policies that support multilingualism can help sustain the visibility of local food traditions. Countries like Belgium, Switzerland, and South Africa have implemented language policies that promote linguistic diversity, which can indirectly benefit the preservation of culinary heritage (Deprez & Plessis, 2000). These policies ensure that regional dialects and languages continue to be valued, allowing food-related terminologies to thrive alongside mainstream languages.

5.3 Example in different results with ChatGPT

https://chatgpt.com/share/67ceb278-9e64-8001-89c3-8c69b17c250b (dutch)

https://chatgpt.com/share/67ceb349-ce94-8001-9293-1c8ebee18c64 (english)

5.4 Example in different results with Google

5.5 Core Problem

As global tourism continues to expand, the need for inclusive, culturally sensitive, and technologically advanced tools becomes increasingly critical. The challenges faced by travelers—ranging from linguistic barriers to the invisibility of local knowledge in mainstream digital platforms—highlight a significant gap in the way information is accessed and presented. Research consistently demonstrates that both search engines and large language models exhibit systemic biases favoring dominant languages, particularly English, which limits the visibility of authentic, community-rooted experiences. This linguistic bias not only distorts travelers’ perceptions but also marginalizes the voices and knowledge systems of less globally dominant regions. Moreover, the deep connection between language and food illustrates how culinary traditions are inextricably linked to local languages and dialects. From the naming of dishes to the transmission of recipes and cultural values, language plays a foundational role in preserving and sharing culinary heritage. However, the decline in dialect usage and the dominance of global languages threaten this connection, risking the loss of unique cultural identities.

6. Objective

The AI-powered app we are developing aims to enhance the travel experience for both domestic and international tourists by addressing key challenges such as language barriers, navigating unfamiliar environments, and finding authentic, local attractions. Based on insights from a user study, which revealed that travelers often struggle to find reliable and up-to-date information—especially when it comes to local food—the app focuses on helping users discover local restaurants and authentic cuisine. This focus on restaurants helps us narrow our scope, providing straightforward and reliable guidance without personalized recommendations.

The user study also highlighted that travelers did not express a significant need for assistance with travel logistics, public transportation, budgeting, or event discovery. Most travelers preferred handling these aspects independently, relying on existing tools like Google Maps for transportation or managing their own budgets. This feedback helped us refine the app’s purpose, ensuring it focuses on providing useful and accurate restaurant recommendations instead. The app integrates real-time data from the Google Search API to ensure recommendations are current and relevant. For primary users, it serves as an interactive guide to explore dining options and connect with the local culinary scene.

We chose the name Starplate for the app, drawing inspiration from the Northern Star as a guiding light for finding food, and combining it with "plate" to represent the local dining experiences at restaurants.

7. App Backend Design

Overview

The backend powers a multilingual, content-aware search engine for finding local restaurants. Instead of using structured business APIs (like Google Places), it extracts organic content such as travel blogs and local reviews, leveraging LLMs, translation, web search, and scraping. The system is built using FastAPI, supports WebSockets, and integrates OpenAI GPT-4o, SerpAPI, and Playwright for rich search and analysis.

Core Workflow

The system follows a multi-step pipeline to deliver high-quality local recommendations:

WebSocket connection is established with the frontend for real-time progress updates.
The user provides a city, country, budget, and optional filters.
A query is constructed like:

  best local restaurants in {city} for {country} food {budget} {filters}

The query is translated to the official local languages using the OpenAI API.
Each translated query is sent to Google Search via SerpAPI, retrieving a list of URLs (blogs, articles, etc.)
The result is parsed using OpenAI's LLM to extract restaurant information, including:

  - name
  - location
  - description
  - reviews (if available)

The cleaned and structured data is returned to the frontend through the WebSocket.

Prompt Strategy

The LLM is instructed to return data in a structured JSON format, using a predefined schema (`RestaurantList`) with the following fields per restaurant:

- local_name: the name in the original language

- translated_english_name: English transliteration or translation

- location: neighborhood or address (optional)

- description: brief English description of the restaurant

- reviews: rating (0–5, optional)

The prompt ensures that the model filters out unrelated content (ads, hotels, generic listings) and only includes valid restaurants that are actually located in the target city.

Technologies Used

FastAPI – Web server with async support
WebSocket – Real-time communication with the frontend
OpenAI GPT-4o-mini – Used for translation, semantic extraction, and reasoning
SerpAPI – Retrieves real Google search results for local queries
Pydantic – Ensures typed and validated data throughout the pipeline

Sample Output Format

json {

 "local_name": "Механа Воденицата",
 "translated_english_name": "Mehana Vodenitsata",
 "location": "Primorski park, Burgas",
 "description": "Traditional Bulgarian tavern offering live music and folk dishes in a rustic setting.",
 "reviews": 4

}

8 App UI Design

When designing an application, the user is the most important factor. The simplest of ways to taking a general user into account in the design process is by considering psychological principles that can affect the ways users experience the app and its functionalities.

Thus, we will continue by outlining important factors based on psychological finding and how they translate in terms of UI design.

8.1 Gestalt laws

The principles of Gestalt (Anik, 2022) (Yee, 2012) outline how people tend to organize the results of their perception. Knowing the ways the average user perceives the content of a screen can be used to design a UI that results in an easier navigation of the app and a more visually appealing interface.

These laws identify the following :

· Symmetry is seen as more organized.

· The mind has the tendency to fill in incomplete information or objects.

· Simple design are good. Simplicity aids in faster understanding of information.

· Elements can be grouped toghether by having a visual connection (same color, same shape, move in the same direction, close position) or a similar arrangement of the elements that creates continuity.

· Similar elements are better grouped to attract the attention of the user.

· Each screen should have a focal point.

· Different colores can be used to define areas that are perceived as separate objects.

· Complex objects tend to be interpreted in the more simple manner.

8.2 Information processing

According to the study by George A. Miller, on average people can remember 7± 2 objects at a time. It was also found that by using certain techniques to aid memorization this limit can be surpassed, but this does not nullify the effect of Miller’s study as UI design should be made as simple to interact with as possible. (Miller, 1994) This implies that each screen of an app should be designed to only have 7±2 elements so users have less to remember and an easier time visually navigating the screen. When it comes to the use of text and images, the “Left-to-Right” theory points that it is more convenient for users to have the most important information on the top left side of the screen. Furthermore, people have 2 visual fields, the right field is responsible for the interpretation of images and the left for text. Putting images on the left and the text on the right would makes it easier for users to process the given information. Further limitations on how people process information is tied to the limitations of motor systems. Multitasking of a motor system or motor systems should be avoided. Focusing on having a task of one motor system introduced at a time makes it easier for users to process the information and to have an easier time paying attention to the task. Thus, screens should not give the users multiple information of different types at the same time. (Yee, 2012)

8.2 Use of Colors

Colors can be used to improve the visual appeal of an app. The only issue is that the color selection is a complex subject. Color schemes can be chosen based on the context they are to be used, different color theories and the desired effect on the user. Nevertheless, it is important that colors are used consistently across the app, and they are not overused. (Anik, 2022) Regarding accessibility, the limited color perception of color-blind people should be considered when choosing contracting colors to highlight different elements. (Anik, 2022) It should also be noted that bright and vivid colors, or a mix of bright and dark colors can tire the eye muscles. This should be avoided to make the app easier to use for longer periods of time. (Yee, 2012)

8.3 Feedback

Feedback is quite important for users to feel that their actions have an effect. (Blair-Early, 2008) It can also affect the ease with which users can remember how to use an app. Feedback should be immediate, consistent, clear, concise and it should fit the context of their actions. (Anik, 2022)

8.4 Navigation

It can be important to make it clear to users how they should begin interacting with an interface. To make it clear it would be useful to make the starting element stand out. This can be done either through a different color, size, hue, shape, orientation etc. (Blair-Early, 2008) Guiding the user can further be done through a visual hierarchy. Important elements should be made to stand out in general due to the phenomenon of inattentional blindness. The Invisible Gorilla Experiment by Simons and Chabris in 1999 showed that people will miss unexpected elements when they are focused on a different task. (Drew, 2013) Another aspect that helps in navigation an interface Is that the user should be able to find a logical consistency in it. The responses to user actions should be consistent and any changes from whatever monotony was created should be predictable. The responses should also be reflective of the content of the interface. (Blair-Early, 2008) Further important aspect to ease navigation for users is providing them with a clear reversal or exit option. Such options give a sense of confidence to users and make navigating the app less stressful once they know that they can opt out if they make a mistake. (Blair-Early,2008)

8.5 Efficiency

Hick’s law states that the more options available the longer it takes to decide. In term of UI design, this implies that menus and navigations systems should be simplifies, either there should be a focus on a few items or elements should be labeled well and similar elements should be grouped together. Another way to decrease options would be to create a visual hierarchy. (Anik, 2022) Fitts Law is based around the connection between target size and distance, and the times it takes to reach a target. When it comes to UI, the law implied that bigger buttons are in general faster to use and better suited for the most frequently used elements. It can reinforce that the steps of a tasks should be contained on the same screen to make navigation more efficient. (Anik, 2022)

9. Branding and Visual Identity

Based on our research into general guiding rules and into the design of the state-of-the-art, we started working on a UI design in Canva.

Screenflow StarPlate

9.1 Color Palette and Font

The design is being made with a softer color scheme, with a stronger shade of blue as its defining color similar to how green is used by TripAdvisor and other travel apps. Blue was chosen for its calming effect and its associations with the sky and open seas, both tied to elegance and a sense of freedom and of exploration. We are also aiming to lean into the idea of tradition, local and beauty to guide the font and the pastel tones of secondary colors in our chosen palette.

Logo StarPlate

As a user enters the app, users can log in or continues as guests and use the app directly. Having an account would give users the ability to bookmark interesting restaurants and foods presented. After moving on from the front screen, users an introductory screen if they are a guest or login for the first time. From there can press either on the introduction screen or the navigation bar at the bottom on a button to change to the searching screen. On the searching screen users have to select the country and region/city they are interested in and then they can choose to view an overview of traditional local foods. From the overview of traditional local foods, users can select a dish they want to try and get a selection of restaurants that sell them. Logged in users can also bookmark dishes. On the search screen users can also select some filters for what kinds of restaurant they are looking for based on distance, price range and star rating. As users get a list of recommended restaurants based on their filters with minimal information of each of them. For further info users can click on the restaurant card from the list and get more information and the location. Logged in users can also bookmark restaurants from each restaurant's individual page.

9.1 Logo Design

Multiple variations have been made with the aim to do some primary research for the feedback of our target user base. The logo was inspired by a compass depiction and the northern stars, both deeply intertwined with the concept of navigation and finding your way in multiple cultures. After deciding on StarPlate as the name of our app, we decided to incorporate a plate in the logos as well to create a cohesive, clear and memorable branding.

10. Final product

Due to time constraints, the app's user interface (UI) is still in the development phase and has not been fully completed yet. While the app is not final in terms of its design and overall presentation, the screenshots and visuals provided below represent the direction we are aiming for. These images illustrate how the results that the algorithm produces look like, showcasing the key feature and layout that will be present once development is finished.

11. User Feedback

We conducted a series of user tests with our prototype, focusing on exploring how it serves different types of users and uncovering specific use cases that could be valuable for further development. The testing led to three particularly interesting and insightful outcomes, which we believe demonstrate the potential of our app to cater to diverse user needs. Here’s a more elaborate explanation of each case:

10.1 Local Experience (User from Varna, Bulgaria)

One user from Bulgaria shared an intriguing experience where the app helped him discover restaurants so hidden and local that even he, as a local resident, wasn’t aware of them. This was significant because the user learned about a restaurant that was somewhat of a "hidden gem" in his city, one that even his own parents had recommended once he asked them. This highlighted an important aspect of the app: it can surface local spots that might not be commonly known or widely advertised, even to people who live in the area.

10.2 Street Food (User from Bengaluru, India)

We were particularly curious about how the app would perform in regions where food culture revolves around street vendors, like in India. One user from a multilingual city in India, where seven different languages are spoken, highlighted a key aspect of local dining culture: in many places, street food is the heart of the culinary scene. Although we initially worried that the app might only show more formal, sit-down restaurants, the results were both surprising and delightful. The app was able to display a wide range of street food vendors, showcasing the true variety of regional dishes that people often enjoy on the streets.

10.3 Small Cities (User from Oosterwolde, Netherlands)

Another user, hailing from Oosterwolde, a small city in Friesland, pointed out an issue that is often overlooked by traditional travel guides: the tendency to ignore smaller cities and towns in favor of major tourist hotspots. Oosterwolde, despite its charm, typically gets overshadowed by larger cities in the region. However, the app was able to uncover valuable suggestions for places to eat in Oosterwolde, showing that it can provide relevant recommendations even in less prominent, smaller locations.

10.4 Conclusion

These user cases were highly valuable in showing us how our app can cater to different needs, from local hidden gems and street food to overlooked small cities. The feedback highlighted not only the app's versatility in addressing diverse food and travel preferences but also the importance of further user testing. There are still more regions and user behaviors to explore, and we believe additional testing will help refine and expand the app’s capabilities, ensuring that it can meet the needs of an even broader audience.

12. Future work and improvements

In conclusion, the AI-powered app has made significant progress in providing real-time, interactive, and personalized assistance for tourists. By addressing language barriers and offering tailored recommendations, the app enhances the overall tourist experience and supports local businesses. While the app’s development is progressing well, there are still some aspects that require further refinement, particularly in terms of linguistic accuracy and coverage.

There are several areas for future work and improvement:

12.1 More Research on Dialects

During our research, we discovered that searching in a dialect has a significant impact on the results provided by the app. However, due to time constraints and a limited sample size, we were unable to fully test this hypothesis. We found that dialects are fluid, difficult to pinpoint, and particularly challenging to translate accurately. As a result, our investigation into dialect-specific interactions was not fully realized during this phase. Despite this, the importance of dialects in language processing is clear, and we plan to revisit this area in future development. Expanding the app’s understanding of regional dialects and incorporating more accurate translations will be a key focus for future updates.

12.2 Improving Linguistic Background

The app’s linguistic framework will be further developed to ensure it offers more precise translations and culturally appropriate responses. This will include additional testing to guarantee that the app can effectively communicate with tourists from various countries, ensuring that no information is lost in translation. During our research, we discovered some linguistic challenges that need further attention. For example, we found that the word for "food" varies across languages. In Dutch, for instance, searching for "eten" versus "voeding" yields different results, highlighting the importance of understanding local language nuances. Additionally, we encountered an issue where "Hollandic" appeared as a language option in Eindhoven, though the algorithm worked well with these inconsistencies. These examples demonstrate that while the app can handle some of these linguistic variations, more linguistic background and in-depth research into translating local languages accurately is needed. This is an area we plan to explore further in future development to improve the app's effectiveness in diverse linguistic contexts.

12.3 Potential Expansion

Looking forward, the app has the potential to expand its scope beyond restaurants to include hidden attractions such as secluded beaches, local events, or other off-the-beaten-path experiences. This will provide tourists with more diverse and unique recommendations, further enriching their travel experience.

11. Planning

The planning for this project is going to be done in a way that is less strict than a traditional way of project planning. We think this is possible due to the short time period in which this project is done, and because the dependence of one task on another in the later stages of the project are rather minimal. To facilitate this process we will use a project management tool called Trello. Trello’s Kanban-style board system allows us to organize tasks into lists, track progress in real-time, and easily adapt to changes as needed. Each task will be categorized based on its status (e.g., "To Do," "In Progress," "Completed"), ensuring transparency and clear accountability within the team.

Week	Tim Hobbes	Eline Smit	Veselin Todorov	Georgi Kalchev	Ania Bărbulescu
1	Initial project setup, Trello organization	Target group analysis, user research	Initial backend setup, language bias research start	Travel app evaluation, problem identification	UX research, user survey draft
3	Technical requirements & planning	Dialects research, literature review	Analysis of LLM and search engines language bias	Exploring Google Search API; Search differences based on language	Questionnaire development, design principles
2	Coordinating app feature list	Continued dialects analysis, target refinement	Integration planning: Google API & LLM	Bias testing in different languages, exploring options for a solution to our problem	UI wireframe design (Canva)
4	First frontend prototype (Java/Android)	State-of-the-art AI travel tools analysis	Backend design: sequence diagram, class diagram	Implemented searching functionality	Logo exploration
5	Initial backend setup	Refined dialect mapping to cuisine & region	LLM integration; defined structured output format (JSON), language detection and query translation	Refined search query logic, optimized Google Search API usage (queries, domains, exact terms)	Conducted user testing, collected feedback
6	Web scraping pipeline, frontend/backend connection	Cleaning the wiki	Connected translated queries with search api, refined LLM interactions	Connected LLM functionality with Google search API, refined search precision	Refined UI based on user feedback
7	Final integration of backend, frontend, and scraping modules	Prepared presentation slides	Finalized LLM-based extraction, restaurant format, translation	Connected query logic with multilingual search; final backend testing	Design summary

12. Bibliography

Ainun, A., Zohriyah, A. M., Sholikhah, U. N., Isnaini, H. N., & Apriyanto, F. R. (2024). Javanese culinary glocalization through learning Indonesian. Proceeding of International Conference Cultures & Languages, 2(1), 563–583. https://doi.org/10.22515/iccl.v2i1.9673

Anik, K., Shahriar, R., Khan, N., & Omi, K. S. I. (2023, March). Elevating software and web interaction to new heights: Applying formal HCI principles for maximum usability. https://doi.org/10.13140/RG.2.2.14304.76803/1

Awah, P. K. (2021). Multilingualism in Cameroon: An expression of many countries in one country. In IntechOpen eBooks. https://doi.org/10.5772/intechopen.99703

Blair-Early, A., & Zender, M. (2008). User interface design principles for interaction design. Design Issues, 24(3), 85–107. http://www.jstor.org/stable/25224185

Deprez, K., & Plessis, T. (2000). Multilingualism and government: Belgium, Luxembourg, Switzerland, former Yugoslavia, South Africa. https://www.semanticscholar.org/paper/Multilingualism-and-government-%3A-Belgium%2C-former-Deprez-Plessis/a7a8abfb601bc516803d04cc75070a4807aad566

Drew, T., Võ, M., & Wolfe, J. (2013, July). The invisible gorilla strikes again: Sustained inattentional blindness in expert observers. Psychological Science, 24. https://doi.org/10.1177/0956797613479386

Luo, Q., Puett, M. J., & Smith, M. D. (2024). A "perspectival" mirror of the elephant: Investigating language bias on Google, ChatGPT, YouTube, and Wikipedia. Queue, 22(1), 23–47.

Miller, G. (1994, April). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 101, 343–352. https://doi.org/10.1037/0033-295X.101.2.343

Muth’im, A., & Sutiono, C. (2024). Maintaining multilingualism in a multi-culture country: The case of Indonesia. Arab World English Journal for Translation and Literary Studies, 8(1), 184–194. https://doi.org/10.24093/awejtls/vol8no1.14

Rovira, C., Codina, L., & Lopezosa, C. (2021). Language bias in the Google Scholar ranking algorithm. Future Internet, 13(2), 31. https://doi.org/10.3390/fi13020031

Segev, E. (2010). Google and the digital divide: The bias of online knowledge. Elsevier.

Sharma, N., Murray, K., & Xiao, Z. (2024). Faux polyglot: A study on information disparity in multilingual large language models. arXiv preprint, arXiv:2407.05502.

Tyas, A. S. P. (2017). Identifikasi kuliner lokal Indonesia dalam pembelajaran bahasa Inggris. Jurnal Pariwisata Terapan, 1(2), 38. https://doi.org/10.22146/jpt.24970

Umiyati, M., & Pratama, N. A. D. Y. (2021). Flora ecolexicon and procedural eco-text of processing Bali local culinary. RETORIKA: Jurnal Ilmu Bahasa, 7(2), 106–114. https://doi.org/10.22225/jr.7.2.3955.106-114

Van Herck, R., & Vermandere, D. (2016). When the language bonus becomes an onus: A Belgian case study. https://www.semanticscholar.org/paper/When-the-language-bonus-becomes-an-onus%3A-a-Belgian-Herck-Vermandere/2b6da3e4f4d7082df8bebb027a2ebb81cecda739

Yee, C., Ling, C., Yee, W., & Zainon, W. M. N. (2012, January). GUI design based on cognitive psychology: Theoretical, empirical and practical approaches. 2, 836–841.

Zurriyati, Z., & Sinar, T. S. (2018). The power of culinary in sustaining Aceh ecolexicon. AICLL Annual International Conference on Language and Literature, 1(1), 75–80. https://doi.org/10.30743/aicll.v1i1.11

14. Appendices

Appendix A - Questions User Study

1. Public transport

a. Prices and cards (weekend card, week card, 48 hours so on)

b. Navigation of public transport

Is often transportation a common necessity in the place you like to visit?
Do you have difficulties finding appropriate (fit within your budged, plans, etc) transportation as a tourist?
How do you usually deal with the issue of transportation while visiting? Local taxi, uber, walking, public transport, rented bikes, rented scooters, rented car, etc?
Where do you find the information on the local transportation?
Do you often need/found it useful to get a card for public transport (weekend card, week card, etc)? Specify in which situations and which kind of card if applicable.

2. Routes for visiting multiple attractions

a. Routing might be too much for a use course – out of the scope

3. Currency information

If applicable, where do you do you find information on the currency of place to visit?
How/ Where do you go to make exchanges of currency, before or after reaching your destination?
Do you have issues finding such information?

4. Food ideas

a. Find the right place for someone

i. Ask for preferences – food preferences, budget, local food

Do you prefer to eat local cuisine or more international cuisine as a tourist?
Do you often have difficulties finding local spots that fit your food preferences and/or restrictions?
If applicable, how do you find places to eat that fit your specific requirements?
Do you plan the food budged in advance? How?
Is there any tool you find useful to find places to fit into the food budged?

ii. Local food summaries

If applicable, where do you get information on the local cuisine?
Would you find a summary of local traditional cuisine and restaurants useful as a tourist?

iii. Avoid tourist places and find actual good local food

Would you want to find local spots to eat instead of more tourists restaurants?
If applicable, how do you find good local spots and attempt to avoid tourist restaurants?

b. Query google and take the data from there

What software do you use to find placed to eat as a tourist? (Google, Yelp, TripAdvisor, etc)?
If applicable, do you find these software useful for finding more traditional local restaurants?

5. Find good not that popular attractions

a. How do they find local hidden gems – for food, attractions,

Do you usually want to visit less known/popular spots?
What kind of hidden gems do you like to visit? Food, events, museums, art installations, graffiti, etc?
How do you find such spots? If you ask someone, who?
If you find them on your own, explain your process of where you look, which kind of sites, blogs, magazines, travel agency, etc.
Do you have difficulties finding such spots?

6. Find events

Do you look for local events as part of the trip?
How do you tend to find out about local events, before and after reaching your destination?
Would you like to attend more international events or more local ones? Take into account possible language barriers you have to deal with.
Do you have difficulties finding local events?

7. Budget planning

If applicable, do you use any technology to plan your trip’s budget?
If yes what do you use and why?
If not, are there any common things that can often make budgeting complicated?
Would you find it useful to have a software to suggest a plan for your trip based on your budged? Why yes or not?

8. Schedule maker

Have you ever used a schedule maker before?
Do you find the option to automatically have your visiting schedule suggested for you useful?
Does a higher level of personalization of the schedule based on your indicated interests make any changes in your opinion?

9. App design

Are there any features that caused you issues or you found annoying in the travel apps that currently exist? If yes, mention the app and explain the issue with its feature.
Are there any features that were very useful to you in the travel apps that currently exist? If yes, mention the app and explain the how its feature was useful.
Are there any features that you haven't (fully) found in existing travel apps but you thought would be very useful? If yes, explain the feature and what it would solve.

Overall questions

Is there some place where you get most of your information when planning?
Do you tend to make a plan in advance or just figure things out on the go?
If on the go, how and what do you use to figure out what to do while visiting?
Have you ever used AI (ChatGPT, etc) to aid in your planning? If yes, how did you find the experience? Any complications?

Appendix B - Literature Research

Number	Summary	Citation
1	Mobile application with p.o.i. and ML to guide tourists between p.o.i.	T. Ghani, N. Jahan, S. H. Ridoy, A. T. Khan, S. Khan and M. M. Khan, "Amar Bangladesh - a Machine Learning Based Smart Tourist Guidance System," 2018 2nd International Conference on Electronics, Materials Engineering & Nano-Technology (IEMENTech), Kolkata, India, 2018, pp. 1-5, doi: 10.1109/IEMENTECH.2018.8465377. keywords: {Smart phones;Mobile applications;Databases;Google;Machine learning;Real-time systems;Linear regression;smartphone;mobile application;technology;machine learning;analysis},
2	Using courd-sourced movement data tourists are guided based on how other tourists went	Basiri, A., Amirian, P., Winstanley, A. et al. Making tourist guidance systems more intelligent, adaptive and personalised using crowd sourced movement data. J Ambient Intell Human Comput 9, 413–427 (2018). https://doi.org/10.1007/s12652-017-0550-0
3	Old system first introducing a tourist guide, without the use of mobile data or GPS	Cheverst, K., Davies, N., Mitchell, K., & Friday, A. (2000). Experiences of developing and deploying a context-aware tourist guide. Proceedings Of The 28th Annual International Conference On Mobile Computing And Networking. https://doi.org/10.1145/345910.345916
4	the a long and windy introduction to the use of API's	El-Sofany, H., & El-Seoud, S. A. (2011). Mobile Tourist Guide â?? An Intelligent Wireless System to Improve Tourism, using Semantic Web. International Journal Of Interactive Mobile Technologies (iJIM), 5(4), 4. https://doi.org/10.3991/ijim.v5i4.1695
5	development of a computer simulator that can evaluate the effectiveness of various evacuation guidance methods for tourists from disaster areas	Emori, N., Izumi, T., & Nakatani, Y. (2016). A support system for developing tourist evacuation guidance. In Springer eBooks (pp. 15–28). https://doi.org/10.1007/978-981-10-0551-0_2
6	Mobile application with p.o.i.	Chavan, R., Bhoir, M., Sapkale, G., & Mhatre, A. (2023). Smart Tourist Guide System. Engpaper Journal.
7	Findings reveal that the guides do not believe that digital development will affect their jobs negatively soon but they know there is a threat of robots with artificial intelligence, the use of digital applications, and smart technologies.	Nazli, M. (2020). THE FUTURE OF TOURIST GUIDANCE CONCERNING THE DIGITAL TECHNOLOGY: a COMPARATIVE STUDY. International Journal of Contemporary Tourism Research, 66–78. https://doi.org/10.30625/ijctr.692463
8	Then by defining collaborative filtering approach on the history meaningful POIs are extracted.	Fenza, G., Fischetti, E., Furno, D., & Loia, V. (2011). A hybrid context aware system for tourist guidance based on collaborative filtering. 2022 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE). https://doi.org/10.1109/fuzzy.2011.6007604
9	The paper also describes our development of an efficient broadcast mechanism which enables visitors' requests for information to be serviced quickly despite the wireless communications infrastructure employed.	Davies, N., Cheverst, K., Mitchell, K., & Friday, A. (1999). “Caches in the air”: disseminating tourist information in the GUIDE system. Proceedings WMCSA’99, 11–19. https://doi.org/10.1109/mcsa.1999.749273
10	This paper is a historical analisis of where the tourist guide comes from	Cohen, E. (1985). The tourist guide. Annals of Tourism Research, 12(1), 5–29. https://doi.org/10.1016/0160-7383(85)90037-4
11	This paper describes mobile tourist guide - a complex system that enables comprehensive up-to date information search along with personalized recommendations and services.	Smirnov, A., Kashevnik, A., Balandin, S. I., & Laizane, S. (2013). Intelligent Mobile Tourist Guide. In Lecture notes in computer science (pp. 94–106). https://doi.org/10.1007/978-3-642-40316-3_9
12	This paper automatically generates p.o.i. by looking at what people are saying on social media	Meehan, K., Lunney, T., Curran, K., & McCaughey, A. (2016). Aggregating social media data with temporal and environmental context for recommendation in a mobile tour guide system. Journal of Hospitality and Tourism Technology, 7(3), 281–299. https://doi.org/10.1108/jhtt-10-2014-0064
13	Our proposed system is a centralized system based on web services which provides all necessary information and tools that can be used by tourists to organize their trip.	Singh, V., Bali, A., Adhikthikar, A., & Chandra, R. (2014). Web and mobile based tourist travel guide system for fiji’s tourism industry. Asia-Pacific World Congress on Computer Science and Engineering, 1–7. https://doi.org/10.1109/apwccse.2014.7053840
14	proposes an alternative network to 4G specially for tour guides	Davies, N., Cheverst, K., Mitchell, K., & Efrat, A. (2001). Using and determining location in a context-sensitive tour guide. Computer, 34(8), 35–41. https://doi.org/10.1109/2.940011
15	evaluating different guideance systems to prevent overcrowding during an emergancy	Kinugasa, S., Izumi, T., & Nakatani, Y. (2012). Evaluation of a support system for large area tourist evacuation guidance: Kyoto simulation results. IEEE, 440–445. https://doi.org/10.1109/scis-isis.2012.6505119
16	The proposed model mimic the human tourism guide, through building relationships between knowledge based-system with the role of tourist-guide	Owaied, H., Farhan, H., Al-Hawamde, N., & Al-Okialy, N. (2011). A model for intelligent tourism guide system. Journal of Applied Sciences, 11(2), 342–347. https://doi.org/10.3923/jas.2011.342.347
17	Explores the pitfalls in a tourist guide system, mainly overcourding	Bornt, Christian & Cheverst, Keith. (2003). Social and technical pitfalls designing a tourist guide system.
18	Google maps with custom p.o.i.	Soe, H., & Sein, M. M. (2017). Tourist Guide Information System using Google Map and GPS. International Journal of Advanced Engineering Research and Science, 4(3), 205–209. https://doi.org/10.22161/ijaers.4.3.32
19	Overall, the feedback emphasizes the need for design improvements to enhance the tourist experience and competitiveness of Yuanjiacun.	Zhang, X. ., Disatapundhu, S. ., & Waijittragum, P. . (2024). AN EXAMINATION OF VISUAL GUIDANCE SYSTEMS FOR TOURIST ATTRACTIONS: CASE STUDY OF YUANJIACUN SCENIC AREA. FOCUS ON ARTS : FAR, SSRU, 2(2), 21–33. retrieved from https://so18.tci-thaijo.org/index.php/forfar/article/view/803
20	A guidance routing system based on GPS, the multi-routes are pre-calculated	US20060100778A1
21	Glasses to capture what the user is looking at for guidance	US10268888B2
22	AR on streetview to show POI	US11692842B2
23	A humanoid looking system to guide users	US11409294B2
24	A guidance system to help tourist using POI on Google Maps	Hema, L., Indumathi, R., Prabhakaran, N., & Kumari, D. (2021). Handheld tourist guidance system using GPS. Materials Today Proceedings, 47, 351–354. https://doi.org/10.1016/j.matpr.2021.04.561
25	Introduction of an E-tourism guide	Smirnov, A., Kashevnik, A., Ponomarev, A., Shchekotov, M., & Kulakov, K. (2015). Application for e-Tourism: Intelligent Mobile Tourist Guide. e-Tourism, 40–45. https://doi.org/10.1109/iiai-aai.2015.190