Source-linked AI summary
Fuzzy Ontology-Based Sentiment Analysis of Transportation and City Feature Reviews for Safe Traveling
Farman Ali, D. Kwak, Pervez Khan, S. M. Riazul Islam, K. H. Kim, K. S. Kwak
TL;DR
Sensor networks and city rating scores provide limited coverage or precision, while reviews and tweets contain useful but difficult-to-process transportation and city-feature opinions. The paper combines fuzzy ontology-based sentiment analysis with SWRL decision-making to determine feature polarity and produce an opinion map. The authors report improved sentiment-analysis precision over an existing system.
Problem
Sensor-based systems may not cover all traffic data, and rating scores and large volumes of reviews do not provide easily extracted, precise city-feature and transportation information.
Source: §1 · p. 2Method
The system retrieves reviews and tweets, extracts feature opinions, and applies fuzzy ontology-based sentiment analysis with SWRL rules to determine transportation and city-feature polarity.
Source: §1 · p. 2, §5.1 · p. 6, §5.2 · p. 7Results
97% average precision was reported for the proposed system, compared with 74.1% for the existing system.
Source: §6 · p. 10Takeaways & Limitations
The system produces transportation and city-feature polarity information and an opinion map intended to help travelers plan routes and ITSs identify traffic problems.
Source: §1 · p. 2, §7 · p. 10
Abstract
from arXiv · showhide
Traffic congestion is rapidly increasing in urban areas, particularly in mega cities. To date, there exist a few sensor network based systems to address this problem. However, these techniques are not suitable enough in terms of monitoring an entire transportation system and delivering emergency services when needed. These techniques require real-time data and intelligent ways to quickly determine traffic activity from useful information. In addition, these existing systems and websites on city transportation and travel rely on rating scores for different factors (e.g., safety, low crime rate, cleanliness, etc.). These rating scores are not efficient enough to deliver precise information, whereas reviews or tweets are significant, because they help travelers and transportation administrators to know about each aspect of the city. However, it is difficult for travelers to read, and for transportation systems to process, all reviews and tweets to obtain expressive sentiments regarding the needs of the city. The optimum solution for this kind of problem is analyzing the information available on social network platforms and performing sentiment analysis. On the other hand, crisp ontology-based frameworks cannot extract blurred information from tweets and reviews; therefore, they produce inadequate results. In this regard, this paper proposes fuzzy ontology-based sentiment analysis and SWRL rule-based decision-making to monitor transportation activities and to make a city- feature polarity map for travelers. This system retrieves reviews and tweets related to city features and transportation activities. The feature opinions are extracted from these retrieved data, and then fuzzy ontology is used to determine the transportation and city-feature polarity. A fuzzy ontology and an intelligent system prototype are developed by using Prot\'eg\'e OWL and Java, respectively.
1. Introduction
Sensor-based monitoring and rating scores leave gaps in precise, timely transportation and city-feature information. The paper proposes fuzzy ontology-based sentiment analysis and SWRL decision-making to extract feature polarity from reviews and tweets.
1. Introduction
- Sensor-based systems may miss traffic-related data and fail to warn drivers before risk zones, while travelers struggle to extract useful guidance from large volumes of reviews.The introduction calls for precise information to identify risky spots and activities.Source: §1 · p. 2
- Reviews and tweets describe transportation issues and city features, but their volume and dispersed locations make useful sentiments difficult for people and ITSs to extract.Examples include traffic collisions, jams, landslides, and street conditions.Source: §1 · p. 2
- Existing systems struggle to identify sentiment strength, and crisp ontologies cannot recover useful information from blurred social-network data.Rating scores also provide insufficiently precise information about city features such as safety, cleanliness, and green space.Source: §1 · p. 2
- The proposed system extracts opinions about transportation activities and city features, uses fuzzy ontology to determine polarity, and applies SWRL rules for decision-making.Its stated contributions include identifying major causes of congestion and negative polarity and producing a city-feature opinion map.Source: §1 · p. 2
2. Related works
Prior work addresses transportation monitoring and sentiment analysis, but the cited review identifies gaps in sensor data coverage and in handling imprecise social-network text. The paper positions fuzzy ontology-based semantic knowledge as a response to these gaps.
2. Related works
- Sensor-network systems and vehicle devices provide limited coverage of traffic-related data, motivating methods to extract useful information from broader sources.The review notes that sensor systems monitor within range and vehicle devices collect short-range information.Source: §2 · pp. 2–3
- Prior social-data sentiment studies address traffic, tourism, transit satisfaction, and other tasks using varied retrieval and classification methods.Examples include rule-based traffic sentiment analysis, naïve Bayes for tourism tweets, and social-media measures of transit safety and cleanliness.Source: §2 · p. 3
- Earlier fuzzy-ontology work monitored six transportation features, while the authors identify gaps in feature coverage, opinion-value assignment, phrase integration, and data retrieval.The paper also argues that single-keyword queries and RDF datasets alone are insufficient for its targeted polarity analysis.Source: §2 · p. 4
- The paper frames its contribution as fuzzy ontology-based semantic knowledge for automatically extracting and processing traffic data that crisp ontologies and sensor systems may not capture.The stated motivation is the difficulty of retrieving imprecise transportation information with those existing approaches.Source: §2 · p. 4
3. The proposed system architecture
The proposed architecture retrieves and filters social-network content, preprocesses it, extracts feature–opinion pairs, and scores sentiment words. Fuzzy ontology and polarity computation are used to determine feature sentiment.
3. The proposed system architecture
- The system retrieves transportation and city-feature content through APIs and queries, then uses a fuzzy ontology-based crawler to follow relevant links from vague sources.The described sources include Twitter, Facebook, news, announcements, and travel sites; REST queries use keywords and operators.Source: §3 · pp. 4–5
- Preprocessing removes symbols and articles, stems words, tags parts of speech, and splits text into clauses containing nouns and verb phrases.These steps prepare sentiment clues and make ontology-based feature tagging easier.Source: §3 · p. 5
- The extraction stage pairs transportation activities and city features with opinion words, validates features against the ontology, and computes feature polarity from opinion phrases.The described cues include nouns as objects and adjectives or adverbs as common sentiment expressions.Source: §3 · p. 5
- SentiWordNet supplies sentiment-word scores for fuzzy-logic polarity computation, assigning zero when a word has no database score.The system combines sentiment words for each feature in a tweet or review sentence.Source: §3 · p. 5
- A positive-and-negative opinion lexicon is also used as a component of sentiment analysis.Source: §3 · p. 5
4. Fuzzy ontology-based semantic knowledge
The paper represents transportation and city-feature knowledge with a fuzzy ontology that can express vague concepts and relations. Feature classes and polarity terms support classification and SWRL-based identification of causes for negative polarity.
4. Fuzzy ontology-based semantic knowledge
- A fuzzy OWL plug-in converts the classic ontology into a fuzzy ontology whose concepts express vague knowledge and uncertainty.The paper states that the classic and fuzzy ontologies share classes, instances, concepts, and axioms, while concept values in the fuzzy ontology are blurry terms.Source: §4 · p. 5
- Feature subclasses such as “jammed,” “slow,” and “traffic collision” can help identify causes associated with strongly negative traffic polarity.The ontology’s subclass structure and polarity terms are used to construct SWRL rules for polarity computation.Source: §4 · p. 6
- The ontology models city features and transportation activities with classes, properties, instances, relationships, and fuzzy data types, then evaluates feature-polarity pairs.The described fuzzy relations can link instances with a degree, such as a park ticket having a high rate at 0.7.Source: §4 · p. 6
5. Sentiment analysis and polarity computation
The section describes a pipeline that filters transportation and city-feature tweets and reviews, extracts feature-linked opinions, and computes polarity with fuzzy ontology and SWRL rules.
5.1. Tweet and review filtration
- Queries with more than 85% recall retrieve candidate posts, which an SVM filters; bigrams and trigrams capture adjacent feature words.A tweet score greater than zero marks it as related to a transportation or city feature; otherwise it is filtered out.Source: §5.1 · p. 6
5.2 Extraction of features and opinion words
- After preprocessing and clause splitting, noun phrases are matched to ontology classes, and feature-specific opinion words are extracted for polarity analysis.The algorithm also tags sentences, identifies subjective and objective opinion text, assigns SentiWordNet scores, and applies an opinion lexicon.Source: §5.2 · p. 7
5.3 SentiWordNet scoring and polarity interval in the ontology
- SentiWordNet supplies opinion-word scores that the ontology maps to strongly positive, positive, neutral, negative, or strongly negative intervals.The stated intervals are strongly positive [0.75-1], positive [0.5-0.75], neutral [0.5], negative [0.25-0.5], and strongly negative [0.0-0.25].Source: §5.3 · p. 7
5.4 Polarity computation based on fuzzy inference layer
- Triangular membership functions and SWRL rules map opinion words across five polarity labels, and defuzzification returns a feature-polarity value.The example computes each rule output as fitness multiplied by input polarity before finding final polarity.Source: §5.4 · p. 8
6. Experiment Results
The experiments compare fuzzy ontology with classic ontology and simple fuzzy logic using retrieved tweets and reviews. Reported results generally favor the fuzzy approach, with results varying for intensively blurred text.
6. Experiment Results
- The evaluation retrieved 1,404 tweet sentences for 7 transportation features and 1,851 review sentences for 13 city features, then compared ontology systems on the same datasets.The systems recorded precision, recall, accuracy, and function measure; the comparison included classic ontology and simple fuzzy logic.Source: §6 · p. 9
- Transportation-feature precision, accuracy, and function measure rose from 73% to 96%, 66% to 89%, and 71% to 94%, respectively, while recall decreased by 0.03%.The authors also report that classic ontology performs better when reviews or tweets are intensively blurred.Source: §6 · p. 9
- Across 13 city features, the proposed system improved over classic ontology by 32% in precision, 24% in accuracy, and 25% in function measure.The comparison is reported in Table 5.Source: §6 · pp. 9–10
- 97% average precision for the proposed system versus 74.1% for the existing system, a reported 23% improvement.The authors describe this as an average precision improvement over Ali et al. (2016).Source: §6 · p. 10
7. Conclusion
The paper concludes that fuzzy-ontology sentiment analysis can determine transportation and city-feature polarity from reviews. It proposes maps and alerts for travelers and identifies extraction improvements and a health-prescription system as future directions.
7. Conclusion
- The framework combines feature extraction, opinion-word extraction, ontology polarity assertions, and fuzzy logic to determine transportation and city-feature polarity.The authors say it can connect with information-retrieval, text-classification, and opinion-mining systems.Source: §7 · p. 10
- The proposed system is intended to provide real-time congestion mapping and city-feature opinion maps, and to notify drivers before risk areas.The conclusion says these functions could save time and fuel and improve transportation safety.Source: §7 · p. 10
- Future work includes improving extraction of true and false opinionated phrases and developing a health-prescription system using disease ontology, neural networks, and semi-supervised filtering.The proposed health system would measure polarity in public-health tweets and suggest drugs in limited time.Source: §7 · p. 10