101-474: Bookish.com is a content discovery and ecommerce website, which launched in February 2013, devoted to books . The site allows users to browse an extensive database of books and authors , add books to user-created digital "shelves", get custom book recommendations, read editorial content and purchase physical books, ebooks , and audiobooks . Bookish was founded in 2011 in a joint venture backed by three of
202-602: A domain hack using .fm , the top level domain of Micronesia , which is popular among FM radio -related sites. The "love" and "ban" buttons enabled users to gradually customize their profiles. Last.fm won the Europrix in 2002 and was nominated for the Prix Ars Electronica in 2003. The Audioscrobbler and Last.fm teams began collaborating closely, moving into the same offices in Whitechapel , London . By 2003, Last.fm
303-452: A "digital bookshelf", was described in a 1990 technical report by Jussi Karlgren at Columbia University, and implemented at scale and worked through in technical reports and publications from 1994 onwards by Jussi Karlgren , then at SICS , and research groups led by Pattie Maes at MIT, Will Hill at Bellcore, and Paul Resnick , also at MIT, whose work with GroupLens was awarded the 2010 ACM Software Systems Award . Montaner provided
404-453: A "similar artists" or "artist fan" radio station. In May 2009, Last.fm introduced Visual Radio, an enhanced version of Last.fm radio. This update brought features such as an artist slideshow and combo stations, which allowed users to listen to stations consisting of common similar artists or up to three artists or three tags. Under the terms of the station's "radio" license, listeners may not select specific tracks (except as previews) or choose
505-641: A citation or recommended article. In such cases, offline evaluations may use implicit measures of effectiveness. For instance, it may be assumed that a recommender system is effective that is able to recommend as many articles as possible that are contained in a research article's reference list. However, this kind of offline evaluations is seen critical by many researchers. For instance, it has been shown that results of offline evaluations have low correlation with results from user studies or A/B tests. A dataset popular for offline evaluation has been shown to contain duplicate data and thus to lead to wrong conclusions in
606-401: A considerable effect beyond the world of scientific publication. In the context of recommender systems a 2019 paper surveyed a small number of hand-picked publications applying deep learning or neural methods to the top-k recommendation problem, published in top conferences (SIGIR, KDD, WWW, RecSys , IJCAI), has shown that on average less than 40% of articles could be reproduced by the authors of
707-447: A dataset that contains information about how users previously rated movies. The effectiveness of recommendation approaches is then measured based on how well a recommendation approach can predict the users' ratings in the dataset. While a rating is an explicit expression of whether a user liked a movie, such information is not available in all domains. For instance, in the domain of citation recommender systems, users typically do not rate
808-516: A few. As the information generated is largely compiled from the ID3 data from audio files "scrobbled" from users' own computers, inaccuracies and misspellings can lead to numerous errors in the listings. Tracks with ambiguous punctuation are particularly prone to separate listings, which can dilute the apparent popularity of a track. Artists or bands with the same name are not always differentiated. The system attempts to consolidate different artist tags into
909-505: A hindrance to lesser-known and unsigned artists' ability to gain exposure for their music, as well as to the overall enjoyment of the site. A new "Play direct from artist" feature was introduced shortly thereafter, allowing artists to select individual tracks for users to stream in full. The ability to listen to custom radio stations, such as "personal tag radio" and "loved tracks radio," was withdrawn on 17 November 2010. This change provoked an angry response among users. Last.fm stated that
1010-450: A hybrid approach, combining collaborative filtering , content-based filtering, and other approaches. There is no reason why several different techniques of the same type could not be hybridized. Hybrid approaches can be implemented in several ways: by making content-based and collaborative-based predictions separately and then combining them; by adding content-based capabilities to a collaborative-based approach (and vice versa); or by unifying
1111-489: A list of pickup points along a route, with the goal of optimizing occupancy times and profits. One of the events that energized research in recommender systems was the Netflix Prize . From 2006 to 2009, Netflix sponsored a competition, offering a grand prize of $ 1,000,000 to the team that could take an offered dataset of over 100 million movie ratings and return recommendations that were 10% more accurate than those offered by
SECTION 10
#17328851882121212-448: A main artist page. This page displays details such as the total number of plays, the total number of listeners, the most popular weekly and overall tracks, the top weekly listeners, a list of similar artists, the most popular tags, and a shoutbox for messages. Additionally, there are links to events, album and individual track pages, and similar artists radio. Official music videos and other videos imported from YouTube may also be viewed on
1313-419: A model from a user's behavior, a distinction is often made between explicit and implicit forms of data collection . Examples of explicit data collection include the following: Examples of implicit data collection include the following: Collaborative filtering approaches often suffer from three problems: cold start , scalability, and sparsity. One of the most famous examples of collaborative filtering
1414-672: A model from a user's past behavior (items previously purchased or selected and/or numerical ratings given to those items) as well as similar decisions made by other users. This model is then used to predict items (or ratings for items) that the user may have an interest in. Content-based filtering approaches utilize a series of discrete, pre-tagged characteristics of an item in order to recommend additional items with similar properties. The differences between collaborative and content-based filtering can be demonstrated by comparing two early music recommender systems, Last.fm and Pandora Radio . Each type of system has its strengths and weaknesses. In
1515-584: A music player application on a computer or an iPod with an Audioscrobbler plugin , or by listening to the Last.fm Internet radio service, either through the Last.fm client or the embedded player. All songs played are added to a log from which personal top artist/track bar charts and musical recommendations are calculated . Last.fm automatically generates a profile page for every user, which includes basic information such as their username, avatar, date of registration, and
1616-406: A new software application for playing Last.fm radio streams and logging tracks played with other media players. Other changes included improvements to the friends system, updating it to require a two-way friendship, the addition of the Last.fm "Dashboard" where users can view relevant information for their profiles on a single page, expanded options for purchasing music from online retailers , and
1717-469: A new visual design for the website (including an optional black colour scheme ). The site began expanding its language offerings on 15 July 2006, starting with a Japanese version. Currently, the site is available in German , Spanish , French , Italian , Polish , Portuguese , Swedish , Russian , Turkish , and Simplified Chinese . In late 2006, Last.fm won the award for Best Community Music Site at
1818-414: A new album's release may continue to be reflected in play data for many months or even years after it drops out of commercial charts. For example, The Beatles have consistently ranked among the top five bands on Last.fm, reflecting the enduring popularity of their music regardless of current album sales. Significant events, such as the release of a highly anticipated album or the death of an artist, can have
1919-551: A potentially overwhelming number of items that a service may offer. Typically, the suggestions refer to various decision-making processes , such as what product to purchase, what music to listen to, or what online news to read. Recommender systems are used in a variety of areas, with commonly recognised examples taking the form of playlist generators for video and music services, product recommenders for online stores, or content recommenders for social media platforms and open web content recommenders. These systems can operate using
2020-558: A proxy that allows users to utilize a media player of their choice. On 24 March 2009, Last.fm announced that access to Last.fm Radio would require a subscription of €3.00 per month for users residing outside the US, UK, and Germany. This change was initially set to take effect on 30 March, but was postponed until 22 April. This decision resulted in over 1,000 comments on the Last.fm blog, most of which were negative. Streaming and radio services were discontinued by Last.fm on 28 April 2014, allowing
2121-411: A rating history similar to the current user or item, they generate recommendations using this neighborhood. Collaborative filtering methods are classified as memory-based and model-based. A well-known example of memory-based approaches is the user-based algorithm, while that of model-based approaches is matrix factorization (recommender systems) . A key advantage of the collaborative filtering approach
SECTION 20
#17328851882122222-502: A separate development-oriented site on 5 September 2005. At the bottom of each Last.fm page, there was an Audioscrobbler " slogan " that changed each time the page was refreshed. Based on well-known sayings or advertisements, these slogans originally appeared at the top of the Audioscrobbler website pages and were created and contributed by the original site members. An update to the site was implemented on 14 July 2006, which introduced
2323-468: A single artist profile and has recently made efforts to harmonize track names. Last.fm generates weekly "global" charts of the top 400 artists and tracks listened to by all Last.fm users. The results differ significantly from traditional commercial music charts provided by the UK Top 40 , Billboard , Soundscan , and others, which are based on radio plays or sales. Last.fm charts are less volatile, and
2424-522: A single criterion value, the overall preference of user u for the item i, these systems try to predict a rating for unexplored items of u by exploiting preference information on multiple criteria that affect this overall preference value. Several researchers approach MCRS as a multi-criteria decision making (MCDM) problem, and apply MCDM methods and techniques to implement MCRS systems. See this chapter for an extended introduction. The majority of existing approaches to recommender systems focus on recommending
2525-922: A single type of input, like music, or multiple inputs within and across platforms like news, books and search queries. There are also popular recommender systems for specific topics like restaurants and online dating . Recommender systems have also been developed to explore research articles and experts, collaborators, and financial services. A content discovery platform is an implemented software recommendation platform which uses recommender system tools. It utilizes user metadata in order to discover and recommend appropriate content, whilst reducing ongoing maintenance and development costs. A content discovery platform delivers personalized content to websites , mobile devices and set-top boxes . A large range of content discovery platforms currently exist for various forms of content ranging from news articles and academic journal articles to television. As operators compete to be
2626-402: A statement, the site indicated that the decision was made to "focus on improving scrobbling and recommendations". On 15 April 2015, Last.fm released a subscriber-exclusive beta version of a new website redesign. Digital Spy described user reactions on the site's forums during the week of the redesign as "universally negative". In 2016, Music Manager was discontinued, and music uploaded to
2727-936: A subscription of €3.00 per month." This change took effect on 22 April 2009. The announcement sparked a wave of disappointment among users, leading to a decline in data submissions, refusal to update signatures or avatars, and even account deletions. On 11 September 2009, CBS Radio announced that Last.fm programming would be available for the first time on four major market FM stations through their HD Radio multicasts. This included KCBS-HD2 in Los Angeles , KITS-HD3 in San Francisco , WWFS-HD2 in New York City , and WXRT-HD3 in Chicago . The programming, which primarily featured music aggregated from Last.fm's user-generated weekly music charts, as well as live performances and interviews from
2828-539: A substantial impact on the charts. The Global Tag Chart displays the 100 most popular tags used to describe artists, albums, and tracks. This is based on the total number of times the tag has been applied by Last.fm users since the tagging system was first introduced and does not necessarily reflect the number of users currently listening to any of the related "global tag radio" stations. Last.fm previously offered customized virtual "radio stations" consisting of uninterrupted audio streams of individual tracks selected from
2929-581: A transplantation problem – recommendations may not apply in all regions (for instance, it would be unwise to recommend a recipe in an area where all of the ingredients may not be available). One example of a mobile recommender system are the approaches taken by companies such as Uber and Lyft to generate driving routes for taxi drivers in a city. This system uses GPS data of the routes that taxi drivers take while working, which includes location (latitude and longitude), time stamps, and operational status (with or without passengers). It uses this data to recommend
3030-473: A useful alternative to search algorithms since they help users discover items they might not have found otherwise. Of note, recommender systems are often implemented using search engines indexing non-traditional data. Recommender systems have been the focus of several granted patents, and there are more than 50 software libraries that support the development of recommender systems including LensKit, RecBole, ReChorus and RecPack. Elaine Rich created
3131-399: Is content-based filtering . Content-based filtering methods are based on a description of the item and a profile of the user's preferences. These methods are best suited to situations where there is known data on an item (name, location, description, etc.), but not on the user. Content-based recommenders treat recommendation as a user-specific classification problem and learn a classifier for
Bookish - Misplaced Pages Continue
3232-477: Is a particularly difficult area of research as mobile data is more complex than data that recommender systems often have to deal with. It is heterogeneous, noisy, requires spatial and temporal auto-correlation, and has validation and generality problems. There are three factors that could affect the mobile recommender systems and the accuracy of prediction results: the context, the recommendation method and privacy. Additionally, mobile recommender systems suffer from
3333-441: Is concerned with finding the most accurate recommendation algorithms. However, there are a number of factors that are also important. Recommender systems are notoriously difficult to evaluate offline, with some researchers claiming that this has led to a reproducibility crisis in recommender systems publications. The topic of reproducibility seems to be a recurrent issue in some Machine Learning publication venues, but does not have
3434-413: Is facing a crisis where a significant number of papers present results that contribute little to collective knowledge [...] often because the research lacks the [...] evaluation to be properly judged and, hence, to provide meaningful contributions." As a consequence, much research about recommender systems can be considered as not reproducible. Hence, operators of recommender systems find little guidance in
3535-463: Is item-to-item collaborative filtering (people who buy x also buy y), an algorithm popularized by Amazon.com 's recommender system. Many social networks originally used collaborative filtering to recommend new friends, groups, and other social connections by examining the network of connections between a user and their friends. Collaborative filtering is still used as part of hybrid systems. Another common approach when designing recommender systems
3636-455: Is only visible to the user and lists suggested new music and events, all tailored to the user's preferences. Recommendations are calculated using a collaborative filtering algorithm , allowing users to browse and hear previews of a list of artists not featured on their own profiles but present on those of others with similar musical tastes. Once an artist has had a track or tracks "scrobbled" by at least one user, Last.fm automatically generates
3737-472: Is substantially improved when blending multiple predictors. Our experience is that most efforts should be concentrated in deriving substantially different approaches, rather than refining a single technique. Consequently, our solution is an ensemble of many methods. Many benefits accrued to the web due to the Netflix project. Some teams have taken their technology and applied it to other markets. Some members from
3838-492: Is that it does not rely on machine analyzable content and therefore it is capable of accurately recommending complex items such as movies without requiring an "understanding" of the item itself. Many algorithms have been used in measuring user similarity or item similarity in recommender systems. For example, the k-nearest neighbor (k-NN) approach and the Pearson Correlation as first implemented by Allen. When building
3939-450: Is the semi-automatic weekly generation and archiving of detailed personal music charts and statistics, which contribute to profile building. Users have access to several different charts, including Top Artists, Top Tracks, and Top Albums. Each of these charts is based on the actual number of listeners for the track, album, or artist, recorded through an Audioscrobbler plugin or the Last.fm radio stream. Additionally, charts are available for
4040-668: The BT Digital Music Awards held in October. Last.fm also partnered with EMI on the Tuneglue-Audiomap project. In January 2007, Last.fm was nominated for Best Website at the NME Awards . At the end of April 2007, rumours surfaced regarding negotiations between CBS and Last.fm, indicating that CBS intended to acquire Last.fm for approximately £225 million ($ 449 million equivalent to $ 635,800,000 in 2023). In May 2007, it
4141-598: The Creative Commons Attribution Share-Alike License and the GNU Free Documentation License . Last.fm currently cannot disambiguate artists with the same name; a single artist profile is shared between valid artists with identical names. Additionally, Last.fm and its users do not differentiate between the composer and the artist of music, which can lead to confusion in classical music genres. One notable feature of Last.fm
Bookish - Misplaced Pages Continue
4242-495: The Federal Trade Commission , led to the cancellation of a second Netflix Prize competition in 2010. Evaluation is important in assessing the effectiveness of recommendation algorithms. To measure the effectiveness of recommender systems, and compare different approaches, three types of evaluations are available: user studies, online evaluations (A/B tests) , and offline evaluations. The commonly used metrics are
4343-455: The big six publishing companies – Hachette Book Group , Penguin Group (USA), and Simon & Schuster – with the goal of increasing the presence of book publishers in the book-buying industry (which was becoming increasingly dominated by Amazon.com due to the increased popularity of online bookstores), as well as to expand the overall book-buying market. The site was expected to launch in
4444-472: The mean squared error and root mean squared error , the latter having been used in the Netflix Prize. The information retrieval metrics such as precision and recall or DCG are useful to assess the quality of a recommendation method. Diversity, novelty, and coverage are also considered as important aspects in evaluation. However, many of the classic evaluation measures are highly criticized. Evaluating
4545-460: The Last.fm studios in New York City, debuted on 5 October. On 12 April 2010, Last.fm announced the removal of the option to preview entire tracks, redirecting users instead to sites such as the free Hype Machine and the pay-to-listen service MOG for this purpose. This decision provoked a significant negative reaction from some members of the Last.fm user community, who perceived the removal as
4646-685: The University of Texas were able to identify individual users by matching the data sets with film ratings on the Internet Movie Database (IMDb) . As a result, in December 2009, an anonymous Netflix user sued Netflix in Doe v. Netflix, alleging that Netflix had violated United States fair trade laws and the Video Privacy Protection Act by releasing the datasets. This, as well as concerns from
4747-417: The above example, Last.fm requires a large amount of information about a user to make accurate recommendations. This is an example of the cold start problem, and is common in collaborative filtering systems. Whereas Pandora needs very little information to start, it is far more limited in scope (for example, it can only make recommendations that are similar to the original seed). Recommender systems are
4848-412: The approaches into one model. Several studies that empirically compared the performance of the hybrid with the pure collaborative and content-based methods and demonstrated that the hybrid methods can provide more accurate recommendations than pure approaches. These methods can also be used to overcome some of the common problems in recommender systems such as cold start and the sparsity problem, as well as
4949-478: The average values of the rated item vector while other sophisticated methods use machine learning techniques such as Bayesian Classifiers , cluster analysis , decision trees , and artificial neural networks in order to estimate the probability that the user is going to like the item. A key issue with content-based filtering is whether the system can learn user preferences from users' actions regarding one content source and use them across other content types. When
5050-460: The breach, along with its connection to similar attacks against Tumblr , LinkedIn , and Myspace during the same timeframe, was not confirmed until August 2016. The passwords were encrypted using an outdated, unsalted MD5 hash . Last.fm informed users of the attack in June 2012. On 14 February 2012, Last.fm announced the launch of a new beta desktop client for public testing. The new scrobbler
5151-553: The communication platform Discord . Last.fm celebrated its twentieth anniversary in 2022. Third-party developers have created programs that integrate users' listening statistical data with Discord, including a popular bot from the Netherlands that has over 400,000 total users. Last.fm Ltd is funded through the sale of online advertising space and monthly user subscriptions. In 2004, the company received its first round of angel money from Peter Gardner, an investment banker who
SECTION 50
#17328851882125252-503: The company in the summer of 2009. The free user account provides access to all the main features listed below. Registered users are also able to send and receive private messages. The newly launched Last.fm Pro user account adds additional features to the free tier, the most notable being the ability to change usernames and gain early access to new features. A Last.fm user can build a musical profile using any or all of several methods: by listening to their personal music collection on
5353-496: The company's existing recommender system. This competition energized the search for new and more accurate algorithms. On 21 September 2009, the grand prize of US$ 1,000,000 was given to the BellKor's Pragmatic Chaos team using tiebreaking rules. The most accurate algorithm in 2007 used an ensemble method of 107 different algorithmic approaches, blended into a single prediction. As stated by the winners, Bell et al.: Predictive accuracy
5454-420: The current research for answering the question, which recommendation approaches to use in a recommender systems. Said and Bellogín conducted a study of papers published in the field, as well as benchmarked some of the most popular frameworks for recommendation and found large inconsistencies in results, even when the same algorithms and data sets were used. Some researchers demonstrated that minor variations in
5555-556: The data over to the RIAA." This led to several public statements from both Last.fm and TechCrunch , with Last.fm denying that it had shared any personal data with the RIAA. The request was reportedly prompted by the leak of U2 's then-unreleased album No Line on the Horizon and its subsequent widespread distribution through peer-to-peer file sharing services such as BitTorrent . Three months later, on 22 May 2009, TechCrunch reported that it
5656-423: The degree to which it has incorporated the risk into the recommendation process. One option to manage this issue is DRARS , a system which models the context-aware recommendation as a bandit problem . This system combines a content-based technique and a contextual bandit algorithm. Mobile recommender systems make use of internet-accessing smartphones to offer personalized, context-sensitive recommendations. This
5757-414: The design of recommender systems that has wide use is collaborative filtering . Collaborative filtering is based on the assumption that people who agreed in the past will agree in the future, and that they will like similar kinds of items as they liked in the past. The system generates recommendations using only information about rating profiles for different users or items. By locating peer users/items with
5858-543: The desktop client would require a subscription in the US, UK, and Germany, although the website radio would remain free in those countries. In January 2014, the website announced on-demand integration with Spotify and introduced a new YouTube -powered radio player. With the introduction of the YouTube player, the standard radio service became a subscriber-only feature. On 26 March 2014, Last.fm announced that it would discontinue its streaming radio service on 28 April 2014. In
5959-516: The development and use of recommendation frameworks, and (7) establish best-practice guidelines for recommender-systems research." Last.fm Last.fm is a music website founded in the United Kingdom in 2002. Utilizing a music recommender system known as "Audioscrobbler," Last.fm creates a detailed profile of each user's musical preferences by recording the details of the tracks they listen to, whether from Internet radio stations or from
6060-474: The evaluation of algorithms. Often, results of so-called offline evaluations do not correlate with actually assessed user-satisfaction. This is probably because offline training is highly biased toward the highly reachable items, and offline testing data is highly influenced by the outputs of the online recommendation module. Researchers have concluded that the results of offline evaluations should be viewed critically. Typically, research on recommender systems
6161-692: The first plugins and subsequently opened an API to the community, which led to support for many music players across different operating system platforms. Audioscrobbler was initially limited to tracking which songs its users played on registered computers, enabling charting and collaborative filtering. Last.fm was founded in 2002 by Felix Miller, Martin Stiksel, Michael Breidenbruecker , and Thomas Willomitzer, all hailing from Germany or Austria . Initially established as an Internet radio station and music community site, it utilized similar music profiles to generate dynamic playlists. The site’s name cleverly employs
SECTION 60
#17328851882126262-458: The first overview of recommender systems from an intelligent agent perspective. Adomavicius provided a new, alternate overview of recommender systems. Herlocker provides an additional overview of evaluation techniques for recommender systems, and Beel et al. discussed the problems of offline evaluations. Beel et al. have also provided literature surveys on available research paper recommender systems and existing challenges. One approach to
6363-428: The first recommender system in 1979, called Grundy. She looked for a way to recommend users books they might like. Her idea was to create a system that asks users specific questions and classifies them into classes of preferences, or "stereotypes", depending on their answers. Depending on users' stereotype membership, they would then get recommendations for books they might like. Another early recommender system, called
6464-537: The gateway to home entertainment, personalized television is a key service differentiator. Academic content discovery has recently become another area of interest, with several companies being established to help academic researchers keep up to date with relevant academic content and serendipitously discover new content. Recommender systems usually make use of either or both collaborative filtering and content-based filtering, as well as other systems such as knowledge-based systems . Collaborative filtering approaches build
6565-408: The hybrid system. Content-based recommender systems can also include opinion-based recommender systems. In some cases, users are allowed to leave text reviews or feedback on the items. These user-generated texts are implicit data for the recommender system because they are potentially rich resources of both feature/aspects of the item and users' evaluation/sentiment to the item. Features extracted from
6666-472: The interactions of a user within a session to generate recommendations. Session-based recommender systems are used at YouTube and Amazon. These are particularly useful when history (such as past clicks, purchases) of a user is not available or not relevant in the current user session. Domains, where session-based recommendations are particularly relevant, include video, e-commerce, travel, music and more. Most instances of session-based recommender systems rely on
6767-438: The items in the system, an item presentation algorithm is applied. A widely used algorithm is the tf–idf representation (also called vector space representation). The system creates a content-based profile of users based on a weighted vector of item features. The weights denote the importance of each feature to the user and can be computed from individually rated content vectors using a variety of techniques. Simple approaches use
6868-470: The knowledge engineering bottleneck in knowledge-based approaches. Netflix is a good example of the use of hybrid recommender systems. The website makes recommendations by comparing the watching and searching habits of similar users (i.e., collaborative filtering) as well as by offering movies that share characteristics with films that a user has rated highly (content-based filtering). Some hybridization techniques include: These recommender systems use
6969-416: The most relevant content to users using contextual information, yet do not take into account the risk of disturbing the user with unwanted notifications. It is important to consider the risk of upsetting the user by pushing recommendations in certain circumstances, for instance, during a professional meeting, early morning, or late at night. Therefore, the performance of the recommender system depends in part on
7070-402: The move was due to licensing reasons. The change meant that a tag radio stream would now include all music tagged as such, rather than just that tagged by individual users, effectively broadening the number of tracks that could be streamed under any one tag set. In March 2012, Last.fm was breached by hackers, resulting in the compromise of more than 43 million user accounts. The full extent of
7171-410: The music files in the music library. This service was discontinued on 28 April 2014. Stations could be based on the user's personal profile , the user's "musical neighbours," or the user's "friends." Additionally, stations could be created based on tags, provided enough music was associated with the same tag. Users could also create stations on the fly , and each artist page allowed the selection of
7272-446: The order in which they are played. However, any of the tracks played may be skipped or banned completely. The appropriate royalties are paid to the copyright holders of all streamed audio tracks in accordance with UK law. The radio stream utilizes an MP3 format encoded at 128 kbit/s and 44.1 kHz, which can be played using the in-page Flash player or the downloaded Last.fm client. Community-supported players are also available, along with
7373-462: The performance of a recommendation algorithm on a fixed test dataset will always be extremely challenging as it is impossible to accurately predict the reactions of real users to the recommendations. Hence any metric that computes the effectiveness of an algorithm in offline data will be imprecise. User studies are rather a small scale. A few dozens or hundreds of users are presented recommendations created by different recommendation approaches, and then
7474-464: The recommendation algorithms or scenarios led to strong changes in the effectiveness of a recommender system. They conclude that seven actions are necessary to improve the current situation: "(1) survey other research fields and learn from them, (2) find a common understanding of reproducibility, (3) identify and understand the determinants that affect reproducibility, (4) conduct more comprehensive experiments (5) modernize publication practices, (6) foster
7575-400: The relevant artist and track pages. Users may contribute relevant biographical details and other information to any artist's main page in the form of a wiki. Edits are regularly moderated to prevent vandalism . A photograph of the artist may also be added. If more than one photograph is submitted, the most popular one is chosen by public vote. User-submitted content is licensed for use under
7676-425: The removal of some old ones. However, this redesign was met with dissatisfaction among some users, who complained about the "unappealing and non-user-friendly layout," bugs, and slow performance. Nonetheless, a month after the redesign, a CBS press release credited it with generating a 20% increase in the site's traffic. Last.fm debuted Portishead 's album Third on 21 April 2008, a week before its release. It
7777-431: The sequence of recent interactions within a session without requiring any additional details (historical, demographic) of the user. Techniques for session-based recommendations are mainly based on generative sequential models such as recurrent neural networks , Transformers, and other deep-learning-based approaches. The recommendation problem can be seen as a special instance of a reinforcement learning problem whereby
7878-484: The site by musicians and record labels became inaccessible. After the integration with Spotify, these tracks could still be played and downloaded where the option was available; however, following the change, artists themselves were unable to access their songs in the Last.fm catalog. The website experienced a slight revival during the COVID-19 pandemic , beginning in 2020, linked to its popularity within music communities on
7979-505: The site's sophisticated “algorithmic software” that offers reading suggestions, acquired Bookish. Recommender system A recommender system (RecSys) , or a recommendation system (sometimes replacing system with terms such as platform , engine , or algorithm ), is a subclass of information filtering system that provides suggestions for items that are most pertinent to a particular user. Recommender systems are particularly useful when an individual needs to choose an item from
8080-512: The summer of 2011, but the launch was delayed due to technical issues relating to data compilation, as well as a lawsuit filed by the United States Department of Justice in 2012 against Apple Inc. and five major publishing companies regarding the pricing of ebooks. The site officially launched in February 2013 with the support of sixteen additional publishing companies. In early 2014, online ebook retailer Zola Books , attracted by
8181-455: The survey, with as little as 14% in some conferences. The articles considers a number of potential problems in today's research scholarship and suggests improved scientific practices in that area. More recent work on benchmarking a set of the same methods came to qualitatively very different results whereby neural methods were found to be among the best performing methods. Deep learning and neural methods for recommender systems have been used in
8282-530: The system is limited to recommending content of the same type as the user is already using, the value from the recommendation system is significantly less than when other content types from other services can be recommended. For example, recommending news articles based on news browsing is useful. Still, it would be much more useful when music, videos, products, discussions, etc., from different services, can be recommended based on news browsing. To overcome this, most content-based recommender systems now use some form of
8383-472: The team that finished second place founded Gravity R&D , a recommendation engine that's active in the RecSys community . 4-Tell, Inc. created a Netflix project–derived solution for ecommerce websites. A number of privacy issues arose around the dataset offered by Netflix for the Netflix Prize competition. Although the data sets were anonymized in order to preserve customer privacy, in 2007 two researchers from
8484-403: The top tracks by each artist in the Last.fm system, as well as the top tracks for individual albums (when the tagging information of the audio file is available). Artist profiles also keep track of a short list of Top Fans, calculated using a formula designed to reflect the importance of an artist in a fan's profile, balancing users who listen to hundreds of tracks against those who listen to only
8585-403: The total number of tracks played. There is also a Shoutbox for public messages. Profile pages are visible to all, along with a list of top artists and tracks, as well as the 10 most recently played tracks (which can be expanded). Each user's profile features a 'Taste-o-Meter' that provides a rating of how compatible the user's music taste is. Last.fm includes a personal recommendations page that
8686-858: The user is the environment upon which the agent, the recommendation system acts upon in order to receive a reward, for instance, a click or engagement by the user. One aspect of reinforcement learning that is of particular use in the area of recommender systems is the fact that the models or policies can be learned by providing a reward to the recommendation agent. This is in contrast to traditional learning techniques which rely on supervised learning approaches that are less flexible, reinforcement learning recommendation techniques allow to potentially train models that can be optimized directly on metrics of engagement, and user interest. Multi-criteria recommender systems (MCRS) can be defined as recommender systems that incorporate preference information upon multiple criteria. Instead of developing recommendation techniques based on
8787-442: The user's computer or portable music devices. This information is transferred ("scrobbled") to Last.fm's database via the music player (including, among others, Spotify , Deezer , Tidal , Qobuz , MusicBee , SoundCloud , and Anghami ) or through a plug-in installed in the user's music player . The data is then displayed on the user's profile page and compiled to create reference pages for individual artists. On 30 May 2007, it
8888-498: The user's likes and dislikes based on an item's features. In this system, keywords are used to describe the items, and a user profile is built to indicate the type of item this user likes. In other words, these algorithms try to recommend items similar to those that a user liked in the past or is examining in the present. It does not rely on a user sign-in mechanism to generate this often temporary profile. In particular, various candidate items are compared with items previously rated by
8989-400: The user, and the best-matching items are recommended. This approach has its roots in information retrieval and information filtering research. To create a user profile , the system mostly focuses on two types of information: Basically, these methods use an item profile (i.e., a set of discrete attributes and features) characterizing the item within the system. To abstract the features of
9090-536: The user-generated reviews are improved metadata of items, because as they also reflect aspects of the item like metadata, extracted features are widely concerned by the users. Sentiments extracted from the reviews can be seen as users' rating scores on the corresponding features. Popular approaches of opinion-based recommender system utilize various techniques including text mining , information retrieval , sentiment analysis (see also Multimodal sentiment analysis ) and deep learning . Most recommender systems now use
9191-436: The users judge which recommendations are best. In A/B tests, recommendations are shown to typically thousands of users of a real product, and the recommender system randomly picks at least two different recommendation approaches to generate recommendations. The effectiveness is measured with implicit measures of effectiveness such as conversion rate or click-through rate . Offline evaluations are based on historic data, e.g.
9292-495: The website's forum. On 19 June 2012, Last.fm launched Last.fm Originals, a new website featuring exclusive performances and interviews with various musical artists. On 13 December 2012, it was announced that Last.fm would discontinue its radio service after January 2013 for subscribers in all countries except the United States, United Kingdom, Germany, Canada, Ireland, Australia, New Zealand, and Brazil. Additionally, radio in
9393-614: The winning solutions in several recent recommender system challenges, WSDM, RecSys Challenge . Moreover, neural and deep learning methods are widely used in industry where they are extensively tested. The topic of reproducibility is not new in recommender systems. By 2011, Ekstrand , Konstan , et al. criticized that "it is currently difficult to reproduce and extend recommender systems research results," and that evaluations are "not handled consistently". Konstan and Adomavicius conclude that "the Recommender Systems research community
9494-452: Was CBS, the parent company of Last.fm, that had handed over the data. Last.fm once again denied this allegation, asserting that CBS could not have provided the data without Last.fm's knowledge. On 24 March 2009, Last.fm announced a change in its free streaming policy. According to the blog post, "[...] In the United States, United Kingdom, and Germany, nothing will change. In all other countries, listening to Last.fm Radio will soon require
9595-495: Was acquired by CBS Corporation through its streaming division CBS Interactive , which is now part of Paramount Global , for £140 million (US$ 280 million, equivalent to $ 396,500,000 in 2023). The site previously offered a radio streaming service, which was discontinued on 28 April 2014. The ability to access the extensive catalogue of music stored on the site was later removed entirely and replaced by links to YouTube and Spotify where available. The current Last.fm website
9696-446: Was announced that Channel 4 Radio would broadcast a weekly show called "Worldwide Chart," reflecting the listening habits of Last.fm users worldwide. On 30 May 2007, it was revealed that Last.fm had been acquired by CBS for £140 million, with Last.fm's current management team remaining in place. In July 2008, the "new generation" Last.fm was launched, featuring a completely new layout, color scheme, and several new features, alongside
9797-505: Was developed from two separate sources, Last.fm and Audioscrobbler, which were merged in 2005. Audioscrobbler began as a computer science project by Richard Jones while he was attending the University of Southampton School of Electronics and Computer Science in the United Kingdom . The term scrobbling is defined as the process of finding, processing, and distributing information related to people, music, and other data. Jones developed
9898-402: Was fully integrated with Audioscrobbler profiles, allowing input through either an Audioscrobbler plugin or a Last.fm station. The sites also shared numerous community forums, although some were unique to each site. The original Audioscrobbler site at the audioscrobbler.com domain name was entirely merged into the new Last.fm site on 9 August 2005. Subsequently, Audioscrobbler.net was launched as
9999-542: Was introduced to the founders as early as 2002. A second round was led by Stefan Glaenzer, joined by Joi Ito and Reid Hoffman , who also purchased shares from Michael Breidenbruecker . In 2006, the company secured its first round of venture capital funding from European investors Index Ventures , whose General Partners Neil Rimer and Danny Rimer joined Last.fm's board of directors , which included Felix Miller, Martin Stiksel, and Stefan Glaenzer (chair). Original founders Felix Miller, Martin Stiksel, and Richard Jones left
10100-426: Was made available as a free stream on the website, attracting 327,000 listeners in 24 hours. It was the first time Last.fm made an album available before its release. On 22 February 2009, TechCrunch reported that "[the] RIAA asked social music service Last.fm for data about its users' listening habits to find individuals with unreleased tracks on their computers. And Last.fm, which is owned by CBS, allegedly handed
10201-407: Was subsequently released for all users on 15 January 2013. On 12 July 2012, Last.fm announced a new website redesign that was open to public beta, inviting feedback from users participating in the testing phase. The redesign officially went live for all users on 2 August 2012. While technology websites received the redesign positively, many users expressed dissatisfaction with the changes on
#211788