Spelling suggestions: "subject:"crossdomain"" "subject:"acrossdomains""
11 |
Graph-based approaches for semi-supervised and cross-domain sentiment analysisPonomareva, Natalia January 2014 (has links)
The rapid development of Internet technologies has resulted in a sharp increase in the number of Internet users who create content online. User-generated content often represents people's opinions, thoughts, speculations and sentiments and is a valuable source of information for companies, organisations and individual users. This has led to the emergence of the field of sentiment analysis, which deals with the automatic extraction and classification of sentiments expressed in texts. Sentiment analysis has been intensively researched over the last ten years, but there are still many issues to be addressed. One of the main problems is the lack of labelled data necessary to carry out precise supervised sentiment classification. In response, research has moved towards developing semi-supervised and cross-domain techniques. Semi-supervised approaches still need some labelled data and their effectiveness is largely determined by the amount of these data, whereas cross-domain approaches usually perform poorly if training data are very different from test data. The majority of research on sentiment classification deals with the binary classification problem, although for many practical applications this rather coarse sentiment scale is not sufficient. Therefore, it is crucial to design methods which are able to perform accurate multiclass sentiment classification. The aims of this thesis are to address the problem of limited availability of data in sentiment analysis and to advance research in semi-supervised and cross-domain approaches for sentiment classification, considering both binary and multiclass sentiment scales. We adopt graph-based learning as our main method and explore the most popular and widely used graph-based algorithm, label propagation. We investigate various ways of designing sentiment graphs and propose a new similarity measure which is unsupervised, easy to compute, does not require deep linguistic analysis and, most importantly, provides a good estimate for sentiment similarity as proved by intrinsic and extrinsic evaluations. The main contribution of this thesis is the development and evaluation of a graph-based sentiment analysis system that a) can cope with the challenges of limited data availability by using semi-supervised and cross-domain approaches b) is able to perform multiclass classification and c) achieves highly accurate results which are superior to those of most state-of-the-art semi-supervised and cross-domain systems. We systematically analyse and compare semi-supervised and cross-domain approaches in the graph-based framework and propose recommendations for selecting the most pertinent learning approach given the data available. Our recommendations are based on two domain characteristics, domain similarity and domain complexity, which were shown to have a significant impact on semi-supervised and cross-domain performance.
|
12 |
Trajectories As a Unifying Cross Domain Feature for Surveillance SystemsWan, Yiwen 12 1900 (has links)
Manual video analysis is apparently a tedious task. An efficient solution is of highly importance to automate the process and to assist operators. A major goal of video analysis is understanding and recognizing human activities captured by surveillance cameras, a very challenging problem; the activities can be either individual or interactional among multiple objects. It involves extraction of relevant spatial and temporal information from visual images. Most video analytics systems are constrained by specific environmental situations. Different domains may require different specific knowledge to express characteristics of interesting events. Spatial-temporal trajectories have been utilized to capture motion characteristics of activities. The focus of this dissertation is on how trajectories are utilized in assist in developing video analytic system in the context of surveillance. The research as reported in this dissertation begins real-time highway traffic monitoring and dynamic traffic pattern analysis and in the end generalize the knowledge to event and activity analysis in a broader context. The main contributions are: the use of the graph-theoretic dominant set approach to the classification of traffic trajectories; the ability to first partition the trajectory clusters using entry and exit point awareness to significantly improve the clustering effectiveness and to reduce the computational time and complexity in the on-line processing of new trajectories; A novel tracking method that uses the extended 3-D Hungarian algorithm with a Kalman filter to preserve the smoothness of motion; a novel camera calibration method to determine the second vanishing point with no operator assistance; and a logic reasoning framework together with a new set of context free LLEs which could be utilized across different domains. Additional efforts have been made for three comprehensive surveillance systems together with main contributions mentioned above.
|
13 |
Metáforas e significados na composição musical : quando o compositor revisita e ressignifica seus passosSoares, Carlos Walter Alves January 2016 (has links)
Esta tese apresenta elucidações técnicas e interpretações póscomposicionais de quatro peças compostas durante o doutorado: “Io” e “Calisto”, para piano solo, “Ganimedes” e “Europa”, para dois pianos. A composição musical é abordada em três etapas: a) descrição dos aspectos musicais e extramusicais dos primeiros estágios da composição; b) investigação das características sonoras das peças; c) construção de interpretações ficcionais pós-composicionais, imaginando as luas de Júpiter como ímpeto composicional e interpretativo. O terceiro estágio é realizado na perspectiva da teoria contemporânea das metáforas, tal qual proposta por George Lakoff e Mark Johnson e, posteriormente, por Mark Turner e Gilles Falconnier. Através da apresentação das categorias de metáforas, cruzamentos de mapeamento de domínios, esquemas de imagens, princípio de invariância, espaços conceptuais e mesclagem conceptual, se estabelecem, neste trabalho, conexões entre os conceitos metafóricos e a conceptualização metafórica da música. Isso contribui para a aproximação entre as características marcantes das peças e as cenas/ações propostas como primeira etapa das interpretações pós-composicionais, construídas pelo compositor. O trabalho propõe-se a pensar sobre a atribuição de significados da composição musical em uma perspectiva global, em que elementos extramusicais podem expandir a assimilação do fenômeno musical, bem como contribuir para a pedagogia da composição musical e para a produção de trabalhos colaborativos. / This thesis presents technical elucidations and post-compositional interpretations of four pieces composed during my doctorate degree: "Io" and "Calisto", for solo piano, "Ganimedes" and "Europa" for two pianos. The music composition is approached in three steps: a) description of the musical and extra musical aspects of the first stages of the composition; b) investigation of the sonorous characteristics of the pieces; c) construction of fictional postcompositional interpretations, imagining Jupiter's moons as compositional and interpretative impetus. The third stage is achieved under the perspective of the contemporary theory of metaphor proposed by George Lakoff and Mark Johnson, and posteriorly developed by Mark Turner and Gilles Falconnier. Through the introduction of metaphor categories, the cross-domain mapping, the image schema, the invariance principle, the mental spaces and the conceptual blending, some connections are established between the metaphoric concepts and the metaphoric conceptualization of music. This contributes to the approximation between the main characteristics of the pieces and the scenes/actions proposed in the first step of the post-compositional interpretations. This work proposes a reflection about the attribution of meanings on the musical composition in a global perspective, where extra musical elements can expand the assimilation of the musical phenomenon, as well as to contribute to the pedagogy of musical composition and the production of collaborative works.
|
14 |
Fairness Measurement and Cross-domain Analysis on Stakeholders' Perspectives with Development Difference Between Southern and Northern Taiwan: A Comparative Study of Cijin Seaside Park and Tamshi Fishers¡¦ WharfHuang, Yu-tien 18 June 2012 (has links)
Public policies in a pluralistic society should be able to clarify and cope with the complicated nature of public issues dominated by ¡§multiple parties, multiple values, and subjective judgment.¡¨ (Farkas & Anderson, 1974) Therefore, instead of merely concluding with the outcome, we should also take input into account when evaluating resource allocation or assessing policy effectiveness. In addition, since the public and the government differ in their judgment criteria and indicators, and multiple parties also have different understandings, the controversy of ¡§inequity¡¨ thus arises. In short, to conduct fairness measurement, we should focus on defining the parties as well as the input and outcome indicators.
This research adopted the Public Affairs Management framework to analyze Cijin Seaside Park in Kaohsiung City and Tamshi Fishers¡¦ Wharf in New Taipei City.
We have conducted a fairness measurement and cross-domain analysis to explore the equity perception in three aspects: economic equity, social fairness, and procedure justice. Moreover, we conducted a secondary data analysis, analyzed expert interviews, and generalized the carrier condition and factual judgment of the PAM framework. Furthermore, we explored the value judgment and interpersonal judgment in PAM framework with fairness measurement of the information integration theory. Our theory is based on the equity rule proposed by Anderson, which also emphasizes the individual and interpersonal comparison of individual input and outcome. Our results are demonstrated as follows:
1. In the individual value analysis of economic equity, both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf have over 10 local stores and tourists that fits one-factor rule of the tourist number factor.
2. In the individual value analysis of social fairness, Cijin Seaside Park has 10 local stores that support one-factor rule of the public support factor. However, Cijin Seaside Park has over eight tourists that fit the adding rule of public support factor and actual request factor, whereas Tamshi Fishers¡¦ Wharf has over eight local
stores and tourists that fit the same rule.
3. In individual value analysis of procedure justice, both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf have over 14 local stores and tourists that fit the unequal-weight averaging rule of policy support factor and administrative cooperation factor.
4. In interpersonal comparison of economic equity, the experimental result shows that the local stores and tourists in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf fit the unequal-weight averaging rule of the integration of their tourist number factors. Also, the local stores and tourists in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf fit the unequal-weight averaging rule of the integration of their revenue (consumption) factors. Moreover, the local stores and tourists in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf both fit the adding rule of their respective tourist number factor and revenue (consumption) factor. This result also supports the ¡§rule of inequity integration.¡¨
5. In interpersonal comparison of social fairness, the local stores in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf fit the unequal-weight averaging rule of the public support factor. They also fit the unequal-weight averaging rule of the actual request factor. As for the tourists, those in both two places fit the unequal-weight averaging rule of the public support factor. Also, they fit the unequal-weight averaging rule of the actual request factor.
6. In interpersonal comparison of procedure justice, the local stores in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf fit the unequal-weight averaging rule of policy support factor. They also fit the unequal-weight averaging rule of the administrative cooperation factor. As for the tourists, those in both Cijin Seaside Park and Tamshi Fishers¡¦ Wharf fit the unequal-weight averaging rule of policy support factor. Also, they fit the unequal-weight averaging rule of the administrative cooperation factor.
We continued to conduct a cross-domain analysis to integrate the above research results, as well as the results of secondary data analysis and expert interviews. As a result, we thereby propose related suggestions and generalize the operation steps of fairness measurement.
|
15 |
The Construction and Review of Citizen Governance and Cross-Domain AnalysisLiu, Shu-yu 19 August 2012 (has links)
This study investigates how the public forum cognition of citizen affect their intension to attend public forum given researching methods of Public Affairs Management Integrated Reference Framework (PAM) and Social Development Matrix(SDM). Based on PAM, we analysed the meaning of citizen governance and the role of citizen through real events, and discussed the cognition and psychological weights of our citizen to participate public forums.
Through the analysis methods of literatures, this study summarized three factors that proposed by Ostrom(1990) - Institution, Commitment and Supervision, and then sequentially made the standard measurement divided into two levels by fair/unfair, high/low and valid/invalid. In this study, we focused on citizens who had participated in interactive management and chose Information Integration Theory (IIT) as our research method. Our research discussed that psychological weights of Institution, Commitment and Supervision, which defined three explanatory variables to estimate the public forum consciousness of citizens. By using SPSS and AVERAGE software, we computed the psychological weights of three factors and understood the attitudes of citizens toward public forum, furthermore, we compared the differences among the weights and offered suggestion and strategies of citizen governing in the future.
The results are as following:
1. Compared to participants from traditional conservative society, our participants from highly developed multi-cultural society are more willing to attend public forums.
2. Participants of different backgrounds assess sequentially the importance of three factors by Supervision, Institution and Commitment.
3. Our results showed that the integrated information patterns(model) are all unequal weighted average pattern(model).
4. All three factors of our research are significant on dependent variable in main effects, two-factor interaction and three-factor interaction.
|
16 |
Military Confidence Building Measures Across the Strait, constitution, cognitive and condition of the analysis - Elected representatives in the southern region.Kang, Chia-Hao 24 August 2012 (has links)
In the study, Cross-Domain analysis is used for ¡§Factual judgement, Value judgement and Interpersonal judgement.¡¨ The result of Factual judgement is made by the interviews of seven retired officers, scholars and experts. And it proceeds as ¡§true¡¨ judgement. Value judgement is indicated by in-depth intervuews with 3 representatives to obtain the ¡§good¡¨ judgement. Interpersonal judgement is based on overall strategies of the ¡§interactive management research group¡¨ and combines it with the perpectives of all patrties, as individuals and as groups, to conclude the ¡§beautiful¡¨ judgement. All the judgements are correspomded to the analysis of the analysis of the constitution, cognitive and conditions of the research topics.
In order to have effective controls of all judgement factors, the interview topics are designed from the study of Mainland China1978 reform, and the rapid economic development, the double digit military spending, and the discussion of ¡§China threat theory¡¨, and the discussion by both sides leaders, academia to establish ¡§Military Confidence Building Measures (MCBMs)¡¨ in the PAM Framemork. The 10 steps of Cross-Domain analysis are used to clarify all studied objects and factors in order to obtain the recognition of experts. The practical policy suggestions are expected.
Under PAM Framework, factual judgement perspective, majority of scholars and retired officers agree the necessary to establish ¡§Military Confidence Building Measures (MCBMs)¡¨. However due to the lack of political trust, time is not yet ripe. Retired officers suggest communication through media may solve the obstacles that impacted by the political environment in southern regions. Scholars have proposed that the southern people are affected by their own political inclinaitions, as well as the idea of their supported parties. From non-government exchange to government exchange, the recognition by people can be achieved by understanding the cultural difference and proceed to political and military trust.
In vaule judgement, representatives realized the topic is involved in cross-strait relationship, political and military aspect and the uncertainty of policy. Therefore, they all took a more conservative point of view. However, they all agree that MCBMs should be based on the removal of missiles, and follow by the current policy ¡§economy first, political after¡¨ to establish non-government exchange. Due to state of hostility, it is recommended to start from oversea activities such as sea rescue and anti-piracy project. Work from political trust toward military trust.
In interpersonal judgemant, a NGT and ISM two stages seminar is held by the ¡§interactive management research group¡¨. The participants voted through a weighted method to conclude 12 specific strategies to establish MCBMs. The strategies are summarized as follow:
1. Removed the missile targeting Taiwan from Mainland China to show the premise of MCBMs.
2. Provide National defense information for each party, to promote cross-strait arms information transparence, and eliminate ideological confrontation.
3. The establishment of cross-strait joint rescue mechanism.
4. The establishment of hotline for leader on both side.
5. The establishment of regular exchange visits of high level military decision makers on both sides.
6. The establishment of cross-strait military information interaction and communication platform.
7. The establishment of oversight system for MCBMs, such as supervised NGO consists of globally well-known and remarkable Cninese accepted by both sides.
8. From civil to government organization to held cross-strait military academic exchange and conference.
9. Showing grace by military budget and arm reduction.
10. Clarify the content of MCBMs. Develop cross-strait affairs negotiators.
11. Signing peace treaty to end hostile confrontation.
12.A nuclear, chemical and biological regulation treaty signed by both sides.
In short, use the cross-domain analysis result to study the possibilities and challenges that might happen during the establishment of MCBMs. The following 5 recommendations can be used for future policy execution reference:
1. Strengthen the policy advocacy, public forums and policy discussion to enhance citizens¡¦ identity.
2. Strengthen cross-strait economic and cultural exchanges to relieve hostile condition.
3. Use ¡§1992 Consensus¡¨ as framework to establish political mutual trust.
4. Establish communication platform to eliminate speculations on both sides.
5. Deepen collaborative platform and signed a peace treaty.
|
17 |
Southern Regional scholars and experts to the Cross-Strait negotiating a peace agreement cognitive and AnalysisPi, Shan-Wei 07 September 2012 (has links)
Chinese President Hu Jin-tao at the 2007 congress report for the first time
"cross-strait can signed a peace agreement" the concept formally written to the
Chinese Communist Party official report file, and Taiwan President Ma at the
Presidential Palace on October 17, 2011, presided over the "golden years series fifth
press conference," first emergency relief after the easy things first, the first
post-administration" to promote the principles of cross-strait relations in the next 10
years, should be highly supportive of domestic public opinion, "countries do need
"and" must be in the case of congressional oversight "three premise, on whether the
two sides negotiating a" cross-strait peace agreement "carefully assess" as echoes,
release both hope that the development of cross-strait peace and goodwill case,
seemingly difficult to promote cross-strait political dialogue seems to have a ray of
hope.
But whether from the Mainland Affairs Council, or the Asia-Pacific Peace
Research Foundation aimed at Taiwan island made routine public opinion polls show
that the people of Taiwan in favor of the establishment that is unified, only a minority,
while nearly 77% of people believe that "the People's Republic of China" is a "
country", and 90% of people do not agree with the Chinese Communist Party says
that" Taiwan is the People's Republic of China under the rule of the provinces
"remarks. Which shows that even in the Taiwan people ideas and circumstances, the
people of Taiwan reunification of Taiwan and the Chinese Communist Party there is
still a considerable degree of exclusion, President Ma, In this case, the Government
signed a "cross-strait peace agreement" will likely contact with the Communist
authorities affects the sensitive nerves of the opposition parties and the people of
Taiwan for "the signing of a peace agreement for unified prelude", the island of
Taiwan public opinion tends to turn the key to negotiating a cross-strait peace
agreement can be completed.
Involve a wide spectrum of view of the "cross-strait peace agreement", should
have considerable professional before they can be aware of another political map of
the island of Taiwan generally showed the "North-Blue and South-Green" distribution,
the study to be analyzed from the point of view of the southern region of scholars and
experts cognition and analysis of the two sides negotiating a peace agreement, and to
further assess the impact of the two sides signed a peace agreement within and
external factors, and the introduction of the parties scholars to analyze the possibility
of discussion on the "cross-strait peace agreement" in order to make recommendations
and Thinking the road.
|
18 |
Metáforas e significados na composição musical : quando o compositor revisita e ressignifica seus passosSoares, Carlos Walter Alves January 2016 (has links)
Esta tese apresenta elucidações técnicas e interpretações póscomposicionais de quatro peças compostas durante o doutorado: “Io” e “Calisto”, para piano solo, “Ganimedes” e “Europa”, para dois pianos. A composição musical é abordada em três etapas: a) descrição dos aspectos musicais e extramusicais dos primeiros estágios da composição; b) investigação das características sonoras das peças; c) construção de interpretações ficcionais pós-composicionais, imaginando as luas de Júpiter como ímpeto composicional e interpretativo. O terceiro estágio é realizado na perspectiva da teoria contemporânea das metáforas, tal qual proposta por George Lakoff e Mark Johnson e, posteriormente, por Mark Turner e Gilles Falconnier. Através da apresentação das categorias de metáforas, cruzamentos de mapeamento de domínios, esquemas de imagens, princípio de invariância, espaços conceptuais e mesclagem conceptual, se estabelecem, neste trabalho, conexões entre os conceitos metafóricos e a conceptualização metafórica da música. Isso contribui para a aproximação entre as características marcantes das peças e as cenas/ações propostas como primeira etapa das interpretações pós-composicionais, construídas pelo compositor. O trabalho propõe-se a pensar sobre a atribuição de significados da composição musical em uma perspectiva global, em que elementos extramusicais podem expandir a assimilação do fenômeno musical, bem como contribuir para a pedagogia da composição musical e para a produção de trabalhos colaborativos. / This thesis presents technical elucidations and post-compositional interpretations of four pieces composed during my doctorate degree: "Io" and "Calisto", for solo piano, "Ganimedes" and "Europa" for two pianos. The music composition is approached in three steps: a) description of the musical and extra musical aspects of the first stages of the composition; b) investigation of the sonorous characteristics of the pieces; c) construction of fictional postcompositional interpretations, imagining Jupiter's moons as compositional and interpretative impetus. The third stage is achieved under the perspective of the contemporary theory of metaphor proposed by George Lakoff and Mark Johnson, and posteriorly developed by Mark Turner and Gilles Falconnier. Through the introduction of metaphor categories, the cross-domain mapping, the image schema, the invariance principle, the mental spaces and the conceptual blending, some connections are established between the metaphoric concepts and the metaphoric conceptualization of music. This contributes to the approximation between the main characteristics of the pieces and the scenes/actions proposed in the first step of the post-compositional interpretations. This work proposes a reflection about the attribution of meanings on the musical composition in a global perspective, where extra musical elements can expand the assimilation of the musical phenomenon, as well as to contribute to the pedagogy of musical composition and the production of collaborative works.
|
19 |
Metáforas e significados na composição musical : quando o compositor revisita e ressignifica seus passosSoares, Carlos Walter Alves January 2016 (has links)
Esta tese apresenta elucidações técnicas e interpretações póscomposicionais de quatro peças compostas durante o doutorado: “Io” e “Calisto”, para piano solo, “Ganimedes” e “Europa”, para dois pianos. A composição musical é abordada em três etapas: a) descrição dos aspectos musicais e extramusicais dos primeiros estágios da composição; b) investigação das características sonoras das peças; c) construção de interpretações ficcionais pós-composicionais, imaginando as luas de Júpiter como ímpeto composicional e interpretativo. O terceiro estágio é realizado na perspectiva da teoria contemporânea das metáforas, tal qual proposta por George Lakoff e Mark Johnson e, posteriormente, por Mark Turner e Gilles Falconnier. Através da apresentação das categorias de metáforas, cruzamentos de mapeamento de domínios, esquemas de imagens, princípio de invariância, espaços conceptuais e mesclagem conceptual, se estabelecem, neste trabalho, conexões entre os conceitos metafóricos e a conceptualização metafórica da música. Isso contribui para a aproximação entre as características marcantes das peças e as cenas/ações propostas como primeira etapa das interpretações pós-composicionais, construídas pelo compositor. O trabalho propõe-se a pensar sobre a atribuição de significados da composição musical em uma perspectiva global, em que elementos extramusicais podem expandir a assimilação do fenômeno musical, bem como contribuir para a pedagogia da composição musical e para a produção de trabalhos colaborativos. / This thesis presents technical elucidations and post-compositional interpretations of four pieces composed during my doctorate degree: "Io" and "Calisto", for solo piano, "Ganimedes" and "Europa" for two pianos. The music composition is approached in three steps: a) description of the musical and extra musical aspects of the first stages of the composition; b) investigation of the sonorous characteristics of the pieces; c) construction of fictional postcompositional interpretations, imagining Jupiter's moons as compositional and interpretative impetus. The third stage is achieved under the perspective of the contemporary theory of metaphor proposed by George Lakoff and Mark Johnson, and posteriorly developed by Mark Turner and Gilles Falconnier. Through the introduction of metaphor categories, the cross-domain mapping, the image schema, the invariance principle, the mental spaces and the conceptual blending, some connections are established between the metaphoric concepts and the metaphoric conceptualization of music. This contributes to the approximation between the main characteristics of the pieces and the scenes/actions proposed in the first step of the post-compositional interpretations. This work proposes a reflection about the attribution of meanings on the musical composition in a global perspective, where extra musical elements can expand the assimilation of the musical phenomenon, as well as to contribute to the pedagogy of musical composition and the production of collaborative works.
|
20 |
Collaborative filtering approaches for single-domain and cross-domain recommender systemsParimi, Rohit January 1900 (has links)
Doctor of Philosophy / Computing and Information Sciences / Doina Caragea / Increasing amounts of content on the Web means that users can select from a wide variety of items (i.e., items that concur with their tastes and requirements). The generation of personalized item suggestions to users has become a crucial functionality for many web applications as users benefit from being shown only items of potential interest to them. One popular solution to creating personalized item suggestions to users is recommender systems. Recommender systems can address the item recommendation task by utilizing past user preferences for items captured as either explicit or implicit user feedback.
Numerous collaborative filtering (CF) approaches have been proposed in the literature to address the recommendation problem in the single-domain setting (user preferences from only one domain are used to recommend items). However, increasingly large datasets often prevent experimentation of every approach in order to choose the one that best fits an application domain. The work in this dissertation on the single-domain setting studies two CF algorithms, Adsorption and Matrix Factorization (MF), considered to be state-of-the-art approaches for implicit feedback and suggests that characteristics of a domain (e.g., close connections versus loose connections among users) or characteristics of data available (e.g., density of the feedback matrix) can be useful in selecting the most suitable CF approach to use for a particular recommendation problem. Furthermore, for Adsorption, a neighborhood-based approach, this work studies several ways to construct user neighborhoods based on similarity functions and on community detection approaches, and suggests that domain and data characteristics can also be useful in selecting the neighborhood approach to use for Adsorption. Finally, motivated by the need to decrease computational costs of recommendation algorithms, this work studies the effectiveness of using short-user histories and suggests that short-user histories can successfully replace long-user histories for recommendation tasks.
Although most approaches for recommender systems use user preferences from only one domain, in many applications, user interests span items of various types (e.g., artists and tags). Each recommendation problem (e.g., recommending artists to users or recommending tags to users) can be considered unique domains, and user preferences from several domains can be used to improve accuracy in one domain, an area of research known as cross-domain recommender systems. The work in this dissertation on cross-domain recommender systems investigates several limitations of existing approaches and proposes three novel approaches (two Adsorption-based and one MF-based) to improve recommendation accuracy in one domain by leveraging knowledge from multiple domains with implicit feedback.
The first approach performs aggregation of neighborhoods (WAN) from the source and target domains, and the neighborhoods are used with Adsorption to recommend target items. The second approach performs aggregation of target recommendations (WAR) from Adsorption computed using neighborhoods from the source and target domains. The third approach integrates latent user factors from source domains into the target through a regularized latent factor model (CIMF). Experimental results on six target recommendation tasks from two real-world applications suggest that the proposed approaches effectively improve target recommendation accuracy as compared to single-domain CF approaches and successfully utilize varying amounts of user overlap between source and target domains. Furthermore, under the assumption that tuning may not be possible for large recommendation problems, this work proposes an approach to calculate knowledge aggregation weights based on network alignment for WAN and WAR approaches, and results show the usefulness of the proposed solution. The results also suggest that the WAN and WAR approaches effectively address the cold-start user problem in the target domain.
|
Page generated in 0.033 seconds