It is widely believed that effective prediction of wastewater treatment results (WTR) is conducive to precise control of aeration amount in the wastewater treatment process (WTP). Conventional biochemical mechanism-driven approaches are highly dependent on complicated and redundant model parameters, resulting in low efficiency. Besides, sharp increase in business volume of wastewater treatment requires automatic operation technologies for this purpose. Under this background, researchers started to introduce the idea of data mining to model the WTP, in order to automatically predict WTR given inlet conditions and aeration amount. However, existing data-driven approaches for this purpose focus on modelling of the WTP at independent timestamps, neglecting sequential characteristics of timestamps during the long-term treatment process. To tackle the challenge, in this paper, a novel prediction and control framework through combination of convolutional neural network (CNN) and recurrent neural network (RNN) is proposed for prediction of the WTR. Firstly, the CNN model is utilized to automatically extract the local features of each independent timestamp in the WTP and make them encoded. Next, the RNN model is employed to represent global sequential features of the WTP on the basis of local feature encoding. Finally, we conduct a large number of experiments to verify efficiency and stability of the proposed prediction framework.

/pdf/data-driven-prediction-and-control-of-wastewater-treatment-3r4ug5ph09.pdf

Data-driven prediction and control of wastewater treatment process through the combination of convolutional neural network and recurrent neural network

As people have become more and more connected, there are certain scenarios where items need to be recommended to groups of users rather than individual user, which motivate studies on group recommender systems (GRSs). However, developing GRSs is not an easy task, because a group consists of multiple members with heterogeneous preferences. How to make a trade-off among their preferences remains challenging. Existing works almost aggregate members’ preferences into forms of single values as group profile. However, simple aggregations fail to well reflect comprehensive group profile when it comes to groups with highly conflicting preferences. In this paper, we propose Greption, a novel group recommendation mechanism from the perspective of preference distribution. First, based on preference distributions toward items in training set, a multi-dimensional support vector regression model is established to predict preference distributions toward candidate items. Then, through a modified VIKOR method, we transform the process of selecting items for a group into a multi-criteria decision making process. Furthermore, the Greption is extended to be able to handle data sparsity. Specifically, we propose two heuristic schemes for this purpose. And we present a set of experiments to evaluate the efficiency of the Greption.

A Novel Group Recommendation Mechanism From the Perspective of Preference Distribution

A recommender system or a recommendation system is a subclass of information filtering system which in turn predicts the “preference” or “ratings” which a user would provide to the specified item. Recommender systems are utilized in a variety of areas comprising news, music, movies, books, search queries, social tags, research articles, and products in general. The primary aim of the recommender system is to allow the computers learn automatically without any human intervention or assistance and regulate activities consequently. The existing methods had a lower amount of search result quality and a minimum rate of ranking accuracy. To overcome this issue and to enhance the ranking quality and search result quality a novel recommender system in the multi-cloud with the use of proposed machine learning algorithm. In this proposed work (NPCA-HAC), the social data set are pre-processed to remove the noise and making them pure. Then, the method of feature selection is carried out with the use of principle component analysis method (PCA). The selected features are then clustered with the use of k-means followed by the Hierarchical Agglomerative Clustering algorithm (HAC). These clusters are then ranked by the use of trust ranking algorithm. Finally, the ranked output was evaluated and the performance measure was analyzed which provides the efficient results from the recommender system.

Efficient Machine Learning Model for Movie Recommender Systems Using Multi-Cloud Environment

With ties among people have been much more closer, making recommendations for groups of users became a more general demand, which facilitates the prevalence of group recommender system (GRS). Existing solutions for GRS are mostly established based on preference feedbacks of absolute form such as ratings, yet neglecting that preference assessment criteria are usually heterogeneous among different members. In this paper, we propose GRS-PR, an enhanced group recommender system by exploiting preference relation. First, a preference relation-based multi-variate extreme learning machine model is formulated to predict unknown preference relations in candidate items. Second, on the basis of predicted results, borda voting rule is employed to generate recommendation results from candidate items. In addition, efficiency, parameter sensitivity, and sparsity tolerance of the GRS-PR are evaluated through a set of experiments.

/pdf/an-enhanced-group-recommender-system-by-exploiting-4jqs0woc4d.pdf

An Enhanced Group Recommender System by Exploiting Preference Relation

As online group activities have increased exponentially, the need for group recommender systems has also increased profoundly. The majority of the recommender systems are designed to recommend to user groups using fixed size preferences of the users. This paper examines the importance of flexible size user preferences in group recommender systems. We propose a variable size preference model in group recommendations both by considering the order in the preferences and also without considering the order. We also study the effect of variable size preferences in group recommendations. Experimental results show that our proposed flexible preference model has increased the overall group satisfaction to a great extent.

A study on the role of flexible preferences in group recommendations

Crowdsourcing activities, carrying out large-scale tasks via wisdoms of crowds, are widely used in practice. However, it is hard for users to find tasks that are suitable for them. Thus, many users participate in tasks, and they are not good at or not interested in, and give answers carelessly or randomly. This phenomenon causes heavy astroturfing problem in crowdsourcing systems, which not only hurts the quality of completing tasks, but also influences user experience. Therefore, recommendation mechanisms that can optimize the match between users and tasks are in demand. However, existing studies simply adopt users’ expertise level or interest degree as the key rule for recommendation. They neglect the fact that interest and expertise function jointly, and that interest can sometimes exert reaction force on expertise. Besides, previous studies assume that users’ interest degree is steady, yet ignoring that it is time-varying rather than static in real world. In this paper, we propose IntexCrowd, fine-grained recommendation mechanism through interest-expertise collaborative awareness for crowdsourcing systems, to curb astroturfing problem. First, the IntexCrowd assigns a topic to each task. Then, topic-specific expertise level as well as interest degree of users are estimated according to historical records of tasks. At last, suitable user lists for topic-specific tasks are suggested as recommendation results. And we present a case study and a set of experiments to confirm the validity of IntexCrowd.

Fine-Grained Recommendation Mechanism to Curb Astroturfing in Crowdsourcing Systems

This paper focuses on recommending items to group of users rather than individual users. To model group profile, existing researches almost aggregate preferences of members into a single value, and thus cannot reflect actual group profile of groups with conflicting characteristics. Therefore, we propose a novel group recommender system mechanism. It views group profile as preference distribution, and then models item recommendation process as a multi-criteria decision making process, in order to obtain better recommendation results. Finally, experiments are conducted to verify the proposed approach.

Yunqing Fu

Papers

Fine-Grained Recommendation Mechanism to Curb Astroturfing in Crowdsourcing Systems

A Novel Group Recommendation Mechanism From the Perspective of Preference Distribution

Beyond the Aggregation of Its Members—A Novel Group Recommender System from the Perspective of Preference Distribution