Smart dark pattern detection: Making aware of misleading patterns through
the intended app
Dr.S.Hrushikesava Raju1, Saiyed Faiayaz Waris2 , Dr.S.Adinarayna3, Dr. Vijaya Chandra Jadala4, G. Subba Rao5
1,4Associate Professor, Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Green
Fields, Vaddeswaram, Guntur
2Assistant Professor, Dept. of CSE, Vignan's foundation for Science, Technology and
Research, Vadlamudi, Guntur
3Professor, Department of CSE, Raghu Institute of Technology,Visakhapatnam
5Assistant Professor, Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Green Fields,
Corresponding Author: email@example.com
Abstract: The significance of dark patterns is to mislead the users while browsing the web. There were social m edia places like
linked in, face book, etc where users find the dark patterns whose aim is to steal users valuable information or attract them to involve
in clicking such advertisements. The domain like UI/UX is used to make dark patterns. Although there many kinds in which they trap
the users attention to focus though the advertisements on the websites where users can be trapped and may lose their money. There are
few security concerns are required to detect the dark patterns through an smart dark pattern detection. The intended theme proposed in
this case is designing the app through which browsing could be done where any dark pattern advertisement is identified, that could be
alerted through a dialog box. For detecting a novel dark pattern detection approach is designed and is co nsidered as in -built app
activity. The accuracy and performance are the main factors that judge the intended theme is designed as per the expectations.
Keywords: Dark patterns, App, Detection, Prevent misleading, IOT, and Report generation
Originally, the advertisements in websites would be one way of attracting the users in clicking and leads to different
behavior. This would leads to purchase some products or some activity that would benefit the intended company income.
This term is introduced by Harry Brignull who used in his phd thesis “cognitive science” in 2010. There are many kinds
of dark patterns that are listed and are demonstrated in the following table. According to originator, the kinds of dark
patterns are listed in his website darkpaterns.org website.
Kind of Dark Pattern
Description and its impact
Bait and Switch
It resulted in unforeseen manner when user is about to do action in desired manner. It
provides value to the content and expects a return. Example: It allows UX pin to get ebooks
for an exchange email address.
These are part of regular content, gets attention of user to click. Example: The dafont.com
site consists of alphabets, misleads to click it. The main download is much smaller and less
visible than ZipMac which has nothing to do.
It allows initially as free service (trial) and start charging after trial period is completed.
Example: Course-era is a global learning platform confuses the new users about the
It asks for user email or social media permissions under false believe-ness for their targets to
achieve. Example: LinkedIn where lawsuit cause fine.
It involves a series of steps, where the last step shows unexpected amount in which product
price plus other amenities like tax, and other components are included. Example: Showing
specific amount in advertisement but shows more different amount at the checkout.
It targets to specific place but won’t notice something else is happening. Example: Skype
software update leads to two other applications like Bing search engine as well as MSN as
It leads to not making informed decisions by the retailer and makes hard for user to compare
one item price with other item. Example: LinkedIn gives free trail and never disclose its
premium subscription charges.
It is found on face book where grabs the user attention over certain things and returns their
information publicly available, than specifically intended. Example: The zapier.com posts in
two modes in which one shows English form that every one understood it and other is filled
with legal jargon without reading it.
It throws the user into certain situation that won’t allow to get out from that place. Exam ple:
Time of India in which once job alerts are registered but won’t have a way to stop alerting
messages from the job portal.
Sneak into Basket
It expects purchase of one thing but additional item also added during the purchasing journey
automatically. Example: GoDaddy site shows specific amount for 3 sites but final amount
includes privacy subscription also which is not meant in the purchasing.
It makes a tricky sentence that asks for one thing but intends to do other thing. Example: Sky
is responsible for products and services that you like unless you click on opt-out.
Big Picture effects
It allows give a statement in the view of future or parallel factor details but not focusing on
small details. Example: Information wielded against democracy during Trump contested in
elections, false news boosted clicks and helps to win trump.
Table 1: List of Dark Patterns that demonstrate how to behave
There is a sequence of activities to place in order to grab the users for reaching the their companies turnover and targets:
Fig.1: Sequence of activities in the process of experiencing the dark patterns
The steps to be taken up for achieving the intended theme are provided as below:
1) Open the app designed with loaded signatures, it is customized to add new signature also or remove outdated signature
2) Open the website in the app for stopping the clicking and alerting such advertisement is a dark pattern.
3) Once detected needs to provide prevention or blocking such advertisements so that end user might experience normal
usage but not unintended experience.
4) At last, a report is alerted or posted to authorized users.
2. Literature Review:
As per information provided by many resources, certain studies are to be mentioned in order to create awareness of dark
patterns and how to identify such patterns is a significant task.
With regard of demonstration provided in , , there are ways to exploit the shoppers or users to attract and involve to
do different behavior through their tricky logics. Although there are bans on these in certain countries, but still many
social media sites are randomly involved these without their interference. Also, it describes how to attract the persons in
making decisions based on their Jobs to be done (JTBD). With regard of article mentioned in , the review of dark
patterns are demonstrated, various ways of dark patterns are explained, how to identify such patterns also specified and
this source act like a guide to the dark patterns. The need of dark patterns is to make money based on the companies turn
over and targets. With respect to the demonstration of , this is a motivational study for the researchers where a lot of
products, a lot of websites, are engaged the dark patterns whose aim is to grab the users to do different behavioral
mannerisms. Based on the nature of patterns, those are classified into few categories. The purpose and details of many
dark patterns are explored in this article. In regard of the information given in , the tricks that these dark patterns will
do is to be alerted early to the users. There are certain functions to be known such as program analysis and few machine
learning approaches are helpful in determining the dark patterns. With regard to the data provided in , this is a survey
made on the dark patterns which aim to mislead the user to do something, also categorize those into specific types and
explored such patterns. Also, surveyed the % of users who are aware of these patterns and the % of users that won’t have
aware of these patterns. With regard of source mentioned in , the objective is to make three things like susceptibility,
making victims, and their impact on the users. There is correlation factor which determines the type of pattern and its
impact, how making users as victims are discussed. As per the demonstration of study in , the nearby spatial devices
that people would use would experience the same dark patterns. This kind of proxemic (social) interactions determines
the root cause of having the dark sides and provides the solutions that would minimize the dangers. With respect to
information of , there are types of dark patterns where classified groups fall under categories labeled as pressure,
Force, Obstacles, Sneaking, and Deception. These categories would be further decomposed into other services. As per the
rationality of mentioned information in , this also demonstrates the variety of dark patterns, their behavior over the
users, and also provided the ways of stopping such patterns with steps initiated. As per study mentioned in , there
were differences listed between the common behavior of UX based dark patterns and AI & machine learning based
algorithms which are shown in the below diagram. The significant difference is the former let’s to change the behavior,
impossible to know you are tricked and latter let’s to mislead the work to do, as well as outrage as tricked.
Fig.2: Difference in activities based on AI & machine based vs UX based
With regard of source mentioned in , the book demonstrates the various laws and legal constructs against deceptive
constructs over digital sites w.r.to Harvard journal of law and technology, Vol.34. In the aspect of source mentioned in
, the dark patterns are having varying designs based on platform they are available, policies are based on technical,
political, and security settings and are demonstrated accordingly. In the view of mentioned information in , there is
specific machine learning classifiers are proposed in order to determine the kind of pattern which is not only dark pattern,
may be of anti-pattern or other type. As per the study of , this shows many statistical analysis on the various kinds of
dark patterns in terms of frequency, when they deceptive, and etc. There are scenarios where certain countries imposed a
fine against the some social running sites because of their carelessness towards grabbing the users to perform misleading
behavior. These laws and fines are imposed by European legislative digital forum authority. With regard of , the
impact of dark patterns are analyzed, their unintended task would trick the users, and also proposed techniques to
minimize such patterns. They are also reviewed in a systematical manner. With the aim to use patterns, the source given
in , the objective is to check come text is in large portion of data where it is there within it or not using indexOf()
which is also called onetime look indexing method. It also allows many patterns at a time and report the statistics about
the patterns against the given text. With regard of information in , the novel tree structure is built up for the
incremental database, and helps to identify the popular patterns in those databases. As per the description from , the
crucial criterion is applied over the temporal databases to determine the regularity in the patterns with respect to various
applications such as stock market, market basket and etc. In the aspect of , all the complaints about the crime are
registered in the law office after the violation of the law is identified by the police or by the automatic violation desig n.
With the view of mentioned information from , the household human patterns are evaluated based on data mining
techniques such as clustering, analysis of energy utilization changes. As per the demonstration of , the apriori
algorithm is used in the infection oriented health environment in early detection of diseases. This could be made scalable
using map reduce technique. The unusual data set can be identified and could be removed. As per the source mentioned in
, the sliding window is used where leaves old transactions and includes new transactions. The novel approach applied
is used to determine positive and negative patterns from the vertical format of the database and wont require to scan
multiple times and also not to construct trees. With regard to the information given in , the advancement of mining
over various application oriented web sites in order to extract the frequent and popular patterns. In this, the pit falls of FP-
Growth are overcome using web usage mining for regular patterns.
As per the information specified in , the detection of islanding is done by SVM based approach and determines
required locations using distributed generation system. As per the source mentioned in , the detection of patterns over
the ZNO structures while loss of insertion in the SAW mixtures is done. With regard to the information in , the image
in the video is decomposed in order to capture the license number using consecutive ANPR and RNPR frame works. As
per the article description given in , the efficiency is achieved by storing the frequent item sets whose support count is
same as earlier conventional approach count in a coalesce matrix as a binary content. With regard of information in ,
the FP-growth and apriori are applied in order to find the regular patterns in multi-dimensional database. It uses the extra
time and also lazy pruning method. In the aspect of source mentioned in , the difference between the centre pixel and
its neighbors are estimated and are iterated based on proposed different approaches for patterns on the required
applications. As per the demonstration of , the local edge detection vs the same method w.r.to the color feature would
be extracted and compared based on evaluation measures considered. With regard of description given in ,
discovering the patterns using novel pattern discovery models is to be development over the text databases and are
minimized the drawbacks associated with the existing data mining techniques. With regard of demonstration of , the
video is monitored and color codes are assigned based on the activity behavior. Hence, the detection of activities and
color is marked on the progress bar. As per the source mentioned in , the detection of animals crossing, speed breaker
and making alerts while moving on the road. This guide helps to the driver in making a safe journey. As per role
mentioned in , objects are tracked when they have more weight satisfying the cut off, those objects are spotted and are
caught using IOT. In the view of description mentioned in , the specific objects are tracked and reports are generated
about those scenarios.
In any article discussed above are useful for determining the patterns using data mining techniques, machine learning
technique and Deep learning approaches. Hence, the significance is taken upon the methodology that could detect the
3. Proposed Work:
In this, the modules identified are designing the app, identifying the dark patterns by the app, displaying an icon over the
misleading advertisements and alerting. The designing the app involves sequence of steps like loading the webpage,
scanning the source code of the web page, Based on advertisement and its intension is to be identified and adds the alert
icon over the misleading advertisement. This manner it leads to make aware of these misleading patterns. The objective of
this system is to alert the user because most of users who are browsing the websites are unintentionally trapped and
involved in such traps. Hence, the ER Diagram of detection of dark patterns through the app is demonstrated, as well as
pseudo procedures of the modules are demonstrated.
In the ER diagram of smart dark pattern detection consist of modules where each module’s functionality is achieved in
Fig.3: ER Diagram of Smart dark patterns detection and Reporting
The intended app specifies its inherent activities that include analyzing the loaded web page using predefined and
efficient extraction tool, and reporting about the identified patterns in that web page as a statistical guide to the end user.
The objective of this theme is to make aware of any user about the dark patterns and try to alert openly in a virtual page
which is similar to actual page but with alerted tags.
The pseudo procedure for smart dark patterns is defined as below:
Pseudo_Procedure SDP_Notification_app(website, alertdialog, parser_darkpatterns):
Step1: open the app, register first, and then login
Step2: load the webpage, open that page in virtual crawler
Step3: Call Analysis_page_source_code(code, detecting_signatures_Darkpatt)
Step4: Call alert and report(output_of_Step3)
The pseudo procedure for Analysis_page_source_code(code,detecting_signatures_darkpatt) is defined as below:
Step1: Read the any tag having money as assignment to the value attribute or shopping address as anchor tag value or
social address as value for anchor tag or as link (or) settimer() method (or) bind() method
Step2: If the part_code contains settimer() method (or) bind() method,
alert(“It is a tricky question – May be one of Confrishaping or Scarcity or CountDowns or Nagging or
Step3: else if part_code contains negative force behavior like NO,Go Back or deactivate account or traping for money
alert(“ It is Force Continuity or Force Enrollment”)
Step4: else if part_code contains Not revealing costs or Checking the membership to close
alert(“ It is Roach model or preselection or Hidden Information or Click-Fatigue”)
Step5: else if part_code contains Sports direct magazine or extra products adding into the basket
alert(” It is Sneak into basket or Hidden Subscription or Hidden Costs”)
Step6: else if part_code contains membership status or flashy visual elements or record abnormal behavior or
advertisement that allows to control
alert(“ It is Trick Question or Misdirection or Bait and Switch or Disguised Ads”)
return output_Analysis_page_source_code that consist of detected patterns and its type in a record
The Pseudo procedure of alert and report(output_Analysis_page_source_code) is given below:
Step1: Store the output_Analysis_page_source_code entity in dictionary form
Step2: Find the length of output_Analysis_page_source_code
Step3: For first entity to last entity based on Step2 where I is loop variable
Create a Warning dialog for detected output_Analysis_page_source_code[i]
set the type detected from the output_Analysis _page_source_code(code, detecting_signatures_darkpatt)
until last entity is reached
Step4: Records the click activity over the dark patterns
Step5: Sends a report to the authorized user through the mail.
In the above mentioned pseudo codes, where app is one module where authentication is checked and opening and
accessing the web page securely, the second module analysis_web_page_darkpatterns in which built-in Parser is used to
identify certain keyword texts, social media or shopping address parts, bifurcates such tags into appropriate dark patterns,
and those category of patterns are tracked in a record, and the third module is alert and report would track of such patterns
and such patterns are tagged with dialog windows, and would send a report to the concerned mail.
The below is the flow chart of Smart dark pattern detection:
Fig.3: Flowchart of Smart dark patterns detection
The expected sequence of screens of this intended objective of the proposed system is defined as follows:
Fig.4: Sequence of activities in the initial stage of SDP App
Fig.5: Identification of Monetary dark pattern in the gaming
In case of Fig.5, a report on this page is prepared and alerted as well as posted to the concerned mail.
Fig.6: Report w.r.to Fig.5
Fig.7: Identification of Roach Motel in the Digest Magazine
In case of Fig.7, a report on this page is prepared and alerted as well as posted to the concerned mail.
Fig.8: Report outputted from Fig.7
The below are few snaps of the dark patterns to make aware of misleading the user when they are browsing:
Fig.8: Few example snaps of dark patterns
The accuracy of SDP app is almost cent percentage compared to many existing approaches in detecting variety of dark
patterns which would mislead the user behavior:
Fig.9: Accuracy of detecting the dark patterns against the approaches vs SDP App
In the Fig.8, the traditional approach reads the dark patterns by theoretical awareness or softcopy of the practicing such
patterns by training, the existing apps would detect but not making aware of such advertisement which possess the dark
patterns, and the proposed objective SDP app makes the opened website in a virtual page and tags if any such dark
patterns are detected along with its type so that the user would be in alert in clicking them.
The estimated objective is to alert the user by making aware of the kind of dark pattern in the web page that the user is using. The
alerts would tagged in a virtual webpage when opened and in-built web page parser is used in order to detect such patterns by tags or
the keywords, also separate boundaries within the web page with addresses related to social media or shopping and etc. This is a wa y
of informing the dark patterns so that user would be cautious. Although it is alerted in a virtual page of an opened page in the
designed app, the user activities on these advertisements which possess behavior of dark patterns are recorded, su ch recorded
activities are stored in a separate file and is communicated to the concerned mail for future usage a nd analysis. The a ccuracy is
appreciable when compared to the existing approaches. The advantages of intended ideology are detecting by loading the sa me page
in virtual page, making aware of dialogs for each predicted advertisement, and avoid internet users to be far from those a ds. The
accessing of the page is fast because page opened in virtually. The limitations are if any new signature is found, need to add to the
existing list for further processing.
 Nikole Wintermeier, Dark Patterns Examples in eCommerce: What they are & why to avoid them, June, 2020,
 Nikole Wintermeier, Decision Science & JBTD for Personalization, March, 2021, https://blog.crobox.com/ article/decision-
 Maier, M., Harr, R., Dark Design Patterns: An End-user Perspective, Human Technology, 16(2): 170-199
 ARUNESH MATHUR et al, Dark Patterns at Scale: Findings from a Crawl of 11K Shopping Web sites, Vol. 3, No. CSCW,
Article 81, September, 2019, https://arxiv.org/pdf/1907.07032.pdf
 Chunyang Chen, Dark-pattern Web Detector, 2019, https://supervisorconnect.it.monash.edu/projects/honours/dark-pattern-web-
 Linda Di Geronimo et al, UI Dark Patterns and Where to Find Them: A Study on Mobile Applications and User Perception, CHI
'20: Proceedings of the 2020 CHI Conference on Hum an Factors in Computing Systems, April 2020, Pages 1–14,
 Aditi M. Bhoot et al, Towards the Identification of Dark Patterns: An Analysis Based on End-User Reactions, IndiaHCI 2020:
IndiaHCI '20: Proceedings of the 11th Indian Conference on Human-Computer In teraction, No vember 2020, Pa ges 24 –33,
 Saul Greenberg et al, Dark patterns in proxemic interactions: a critical perspective, DIS '14: Proceedings of the 2014 conference
on Designing interactive systems, June 2014,Pages 523–532, https://doi.org/10.1145/2598510.2598541.
 Dark pattern detection project, https://dapde.de/en/dark-patterns-en/types-and-examples-en/
 Dark Patterns: A New Scientific Look at UX Deception, https://www.fyresite.com/dark-patterns-a-new-scientific-loo k-at-ux-
 Zakary Kinnaird, Dark patterns powered by machine learning: an intelligent combination, October, 2020,
 Lauren E. Willis, DECEPTION BY DESIGN, Harvard Journal of Law & Technology, Volume 34, Number 1 Fall, 2020,
 Caroline sinders, Dark Patterns and Design Policy, May, 2020, https://points.datasociety.net/dark-patterns-and-design-policy-
 ROBERT NORD AND ZACHARY KURTZ, Using Machine Learning to Detect Design Patterns, March, 2020,
 Federico Caruso, Dark patterns: born to mislead, November, 2019, https://www.europeandatajournalism.eu/ eng/News/Data -
 Corina Cara, DARK PATTERNS IN THE MEDIA: A SYSTEMATIC REVIEW, January, 2020, https://www.
 Hrushikesava Raju S. , Nagabhushana Rao M., “Pattern Matching Using Da ta Preprocessing With The Help Of One Tim e Lo ok
Indexing Method”, International Journal of Pharmacy and Technology, Vol.8(3),pp.18395 -184 07,2016, I SSN:0975-766X,
 Vijay Kumar, G., Sreedevi, M., Bhargav, K., & Mohan Krishna, P. (2018), Incremental mining of popular patterns from
transactional databases, International Journal of Engineering and Technology(UAE), 7, 636-641.
 Vijay Kumar, G., Vishnu Sravya, S., & Satish, G. (2018), Mining high utility regular patterns in transactional dat abase,
International Journal of Engineering and Technology(UAE), 7, 900-902.
 Vijay Kumar, G., Sreedevi, M., Vamsi Krishna, G., & Sai Ram, N. (2018), Regular frequent crime pattern mining on crime
datasets, International Journal of Engineering and Technology(UAE), 7, 972-975.
 Akhila, G., Madhubhavana, N., Ramareddy, N. V., Hurshitha, M., & Ravinder, N. (2018), A survey on health prediction using
human activity patterns through smart devices, International Journal of Engineering and Technology(UAE), 7(1), 226-229.
 Bisoyi, S. S., Mishra, P., & Mishra, S. (2018), Extracting global exceptional frequent pattern from distributed data sources: A
MapReduce approach, Journal of Advanced Research in Dynamical and Control Systems, 10(2 Special Issue), 1 460-1467.
 Kumar, N. V. S. P., & Rajasekhara Rao, K. (2018), A sliding window approach to mine negative and positive regular patterns in
incremental databases using vertical data format, International Journal of Engineering and Technology(UAE), 7(3.27 Special Issue
 Nallamala, S. H., Pathuri, S. K., & Koneru, S. V. (2018), An appraisal on recurrent pattern a nalysis a lgorithm from the net
monitor records, International Journal of Engineering and Technology(UAE), 7, 542-545.
 Rao, G. S., & Rao, G. K. (2018), SVM based pattern recognised islanding detection approach in a multiple distributed generation
system, International Journal of Engineering and Technology(UAE), 7(1), 228-231. doi:10.14419/ijet.v7i1.9559
 Sai Krishna Santosh, G., Mahesh Kumar, K., Siva Kumar, K. P. M., Brahma Sai, K., Sra vani, P., & Shanmukh, G. (2018),
Investigation of insertion loss in SAW delay line with periodically-patterned ZnO structure, Journal of Advanced Research in
Dynamical and Control Systems, 10(2), 541-546.
 Sheela Rani, C. M., Dheeraj, K., Sai Vineeth Reddy, P., & Satyasai, K. (2018), Image segmentation for pattern recognition in
surveillance, International Journal of Engineering and Advanced Technology, 7(3), 45-49.
 Sireesha, M., Vemuru, S., & TirumalaRao, S. N. (2018), Coalesce based binary table: An enhanced algorithm for mining
frequent patterns, International Journal of Engineering and Technology(UAE), 7(1.5 Special Issue 5), 51-55.
 Sreedevi, M., Harika, V., Anilkumar, N., & Sai Thriveni, G. (2018), Regular pattern mining on multidim ensional databases,
International Journal of Engineering and Technology(UAE), 7(2), 61-63. doi:10.14419/ijet.v7i2.20.11752.
 Sucharitha, G., & Senapati, R. K. (2018), Local extreme edge binary patterns for face recognition and image retrieval, Journal of
Advanced Research in Dynamical and Control Systems, 10, 644-654.
 Sucharitha, G., & Senapati, R. K. (2018), Local quantized edge binary patterns for colour texture image retrieval, Jo ur nal of
Theoretical and Applied Information Technology, 96(2), 291-303.
 Changala, R., & Rajeswara Rao, D. (2017), A survey on development of pattern evolving model for discovery of patterns in text
mining using data mining techniques, Journal of Theoretical and Applied Information Technology, 95(16), 3974-3981.
 Dr S.Hrushikesava Raju, Dr M.Nagabhushana Rao, Dr N.Sudheer, Dr P.Kavitharani, Quick Identification of Specific Activity by
Processing of Large-Size Videos Using Advanced Spotter, International Journal of Engineering & Technology(UAE), ISSN:
 Dr S.Hrushikesava Raju, Dr M.Nagabhushana Rao, Dr N.Sudheer, Dr P.Kavitharani, Visual Safe Road Travel App Over Google
Maps About the Traffic and External Conditions, International Journal of Engineering & Technology(UAE), ISSN: 2227-524X,
 Radha Mothukuri, Dr.S.Hrushikesava Raju, S.Dorababu and Saiyed Faiayaz Waris, Smart Catcher of weighted Objects Smart
Catcher of weighted Objects, IOP Conference Series Materials Science and Engineering, 981,2,10.1088/1757-
 V. L. Lalitha, S. H. Raju, V. K. Sonti and V. M. Mohan, Customized Smart Object Detection: Statistics of detected objects using
IoT, 2021 International Conference on Artificial Intelligence and Smart Systems (ICAIS), 2021, pp. 1397-1405, doi: