Behavior Sequence(行为序列)研究综述
Behavior Sequence 行为序列 - Recent methods in sequential recommendation focus on learning an overall embedding vector from a user's behavior sequence for the next-item recommendation. [1] Design/methodology/approachFirst, three kinds of interval grey relational operators for the behavior sequence of a dimensionless system are proposed. [2] Additionally, a two-level attention mechanism is developed to capture the user’s preference: 1) User-specific relation-aware attention layer, which captures the user’s fine-grained preferences with different focus on relations for learning outfit representation; 2) Target-aware attention layer, which characterizes the user’s latent diverse interests from his/her behavior sequences for learning user representation. [3] Sequential recommendation is intended to model the dynamic behavior regularity through users' behavior sequences. [4] Therefore, we propose a Deep Multi-Sequences fused neural Networks (DMSN) to predict intention destinations from fused multi-behavior sequences. [5] Sequential recommendation (SR) aims to recommend items based on user information and behavior sequences. [6] Recently, researchers have shown an increased interest in capturing users’ dynamic and evolving fraudulent tendencies from their behavior sequences. [7] We propose to encode the behavior sequence with two corresponding components: the convolutional network for interactive extraction of users' long-term interests and the long-short term gated fusion module for better combination of long-short preferences Our entire model has been test on multiple real-world data sets, and the results demonstrate that our model is more effective than other recent models on multiple evaluation benchmarks. [8] In ADNNet, a convolutional neural network is used to extract the short-term patterns in the behavior sequences, and a gated recurrent unit is used to mine the long-term patterns in the behavior sequences. [9] Originality/value This study adds to the limited literature on the emotional consequences of KH from knowledge hiders’ perspective and unfolds the behavior-emotion-behavior sequence through the emotional pathway. [10] Moreover, there is no framework that unifies recent behavior sequences and reviews. [11] Combined with content, behavior sequence and other methods, UCINET is used for visual analysis and output to find out the main factors that affect interaction behavior and the tendency of learners to choose interaction objects. [12] Most prior works usually obtain an overall representation based on the user’s behavior sequence, which can not sufficiently reflect the multiple interests of the user. [13] In this paper, we propose a research model on video watching behaviors of MOOC learners, by encoding the behavior sequences retrieved from the videos’ log data, evaluating the similarity between different online learners, clustering and visualizing the similarity results. [14] To combine these behavior effectively, we propose a hierarchical attention mechanism, where the bottom attention layer focuses on the inner parts of each behavior sequence while the top attention layer learns the inter-view relations between different behavior sequences. [15] Moreover, few studies treated the time budget of behaviors as compositions and little was done to characterize the distribution of durations of behavior sequences in relation with health. [16] The results also demonstrate that the length of the recent click-through behavior sequence has an important effect on the prediction performance of the model. [17] Despite their effectiveness, we argue that such left-to-right unidirectional models are sub-optimal due to the limitations including: \begin enumerate* [label=series\itshape\alph*\upshape)] \item unidirectional architectures restrict the power of hidden representation in users' behavior sequences; \item they often assume a rigidly ordered sequence which is not always practical. [18] Many of our behavior sequences are no longer aware, because they have become automated mechanisms. [19] They either only consider the relative position of each behavior in the behavior sequence, or process the continuous temporal features into discrete category features for subsequent tasks, which can hardly capture the dynamic preferences of a user. [20] In this paper, we propose to use the powerful Transformer model to capture the sequential signals underlying users' behavior sequences for recommendation in Alibaba. [21] How to capture users' dynamic and evolving interests from their behavior sequences remains a continuous research topic in the CTR prediction. [22] Personalized Markov model was exploited to mine short-term preferences based on individual’s behavior sequences. [23]顺序推荐中的最新方法侧重于从用户的行为序列中学习整体嵌入向量以进行下一项推荐。 [1] 设计/方法/途径首先，提出了用于无量纲系统行为序列的三种区间灰色关系算子。 [2] 此外，开发了一个两级注意机制来捕捉用户的偏好：1）用户特定的关系感知注意层，它捕捉用户的细粒度偏好，不同关注关系用于学习服装表示； 2) Target-aware attention layer，从他/她的行为序列中表征用户潜在的不同兴趣，用于学习用户表示。 [3] 顺序推荐旨在通过用户的行为序列对动态行为规律进行建模。 [4] 因此，我们提出了一种深度多序列融合神经网络（DMSN）来从融合的多行为序列中预测意图目的地。 [5] 顺序推荐（SR）旨在根据用户信息和行为序列推荐物品。 [6] 最近，研究人员对从用户的行为序列中捕捉用户动态和不断演变的欺诈倾向表现出越来越大的兴趣。 [7] 我们建议用两个对应的组件对行为序列进行编码：用于交互式提取用户长期兴趣的卷积网络和用于更好地组合长短期偏好的长短期门控融合模块我们的整个模型已经在多个真实世界的数据集，结果表明我们的模型在多个评估基准上比其他最近的模型更有效。 [8] 在 ADNNet 中，卷积神经网络用于提取行为序列中的短期模式，而门控循环单元用于挖掘行为序列中的长期模式。 [9] 原创性/价值 这项研究从知识隐藏者的角度增加了关于 KH 情绪后果的有限文献，并通过情绪途径展开了行为-情绪-行为序列。 [10] 此外，没有统一最近的行为序列和评论的框架。 [11] 结合内容、行为序列等方法，利用UCINET进行可视化分析和输出，找出影响交互行为的主要因素以及学习者选择交互对象的倾向。 [12] 大多数现有的工作通常根据用户的行为序列来获得一个整体的表示，这不能充分反映用户的多重兴趣。 [13] 在本文中，我们提出了一种关于 MOOC 学习者观看视频行为的研究模型，通过对从视频日志数据中检索到的行为序列进行编码，评估不同在线学习者之间的相似性，对相似性结果进行聚类和可视化。 [14] 为了有效地结合这些行为，我们提出了一种分层注意力机制，其中底层注意力层关注每个行为序列的内部，而顶层注意力层学习不同行为序列之间的视图间关系。 [15] 此外，很少有研究将行为的时间预算视为组合物，并且很少有研究描述与健康相关的行为序列持续时间的分布。 [16] 结果还表明，最近点击行为序列的长度对模型的预测性能有重要影响。 [17] 尽管它们很有效，但我们认为这种从左到右的单向模型是次优的，因为这些限制包括： \begin enumerate* [label=series\itshape\alph*\upshape)] \item 单向架构限制了用户行为序列中的隐藏表示； \item 他们经常假设一个严格有序的序列，这并不总是实用的。 [18] 我们的许多行为序列不再有意识，因为它们已成为自动化机制。 [19] 他们要么只考虑每个行为在行为序列中的相对位置，要么将连续的时间特征处理成离散的类别特征用于后续任务，很难捕捉到用户的动态偏好。 [20] 在本文中，我们建议使用强大的 Transformer 模型来捕获用户行为序列背后的序列信号，以便在阿里巴巴进行推荐。 [21] 如何从用户的行为序列中捕捉用户的动态和不断变化的兴趣仍然是 CTR 预测中的一个持续研究课题。 [22] 利用个性化马尔可夫模型来挖掘基于个人行为序列的短期偏好。 [23]
User Behavior Sequence 用户行为序列
We treat user behavior sequences in batches. [1] We define user behavior sequences for user behavior data and classify user behavior. [2] Using the method of theoretical research and model analysis, the relationship between user behavior sequence and user behavior characteristics is studied from three aspects of data collection, data processing and data application. [3] The user preference information implicits that the user behavior sequence has an important impact on the click-through rate estimation. [4] The sampler model aims to generate new user behavior sequences based on the observed ones, while the anchor model is leveraged to provide the final recommendation list, which is trained based on both observed and generated sequences. [5] BIH adds specific behaviors to the user behavior sequence and adds a behavior attention layer, which can learn the expression of user interests more accurately. [6] Methodologically, user behavior sequences are constructed as graph-structured data, and we apply two similar graph self-attention networks to model the item transitions and the category click probability. [7] In order to identify the abnormal login during each behavior process of authenticated user, we take largest coincident part of the user behavior sequence and short coincide into consideration. [8] Existing studies explored user behavior sequences in sessions in different ways to enhance query suggestion or document ranking. [9] It dynamically fuses and extracts users' current activated core interests from noisy user behavior sequences. [10] In order to identify the abnormal login during each behavior process of authenticated user, we take largest coincident part of the user behavior sequence and short coincide into consideration. [11] Recommendations based on user behavior sequences are becoming more and more common. [12] , identifying the influential individuals and understanding the interaction among user behavior sequences. [13] Things turn to be more difficult when it comes to long sequential user behavior data, as the system latency and storage cost increase approximately linearly with the length of user behavior sequence. [14] With two-decade rapid development, nowadays the cumulated user behavior sequences on mature Internet service platforms have become extremely long since the user's first registration. [15]我们分批处理用户行为序列。 [1] 我们为用户行为数据定义用户行为序列，并对用户行为进行分类。 [2] 运用理论研究和模型分析的方法，从数据采集、数据处理和数据应用三个方面研究用户行为序列与用户行为特征的关系。 [3] 用户偏好信息隐含了用户行为序列对点击率估计有重要影响。 [4] 采样器模型旨在基于观察到的用户行为序列生成新的用户行为序列，而锚定模型用于提供最终推荐列表，该列表基于观察到的序列和生成的序列进行训练。 [5] BIH在用户行为序列中加入了特定的行为，并增加了行为注意层，可以更准确地学习用户兴趣的表达。 [6] 在方法论上，用户行为序列被构建为图结构数据，我们应用两个相似的图自注意力网络来对项目转换和类别点击概率进行建模。 [7] 为了识别认证用户在每个行为过程中的异常登录，我们考虑了用户行为序列的最大重合部分和短重合。 [8] 现有研究以不同方式探索会话中的用户行为序列，以增强查询建议或文档排名。 [9] 它从嘈杂的用户行为序列中动态融合并提取用户当前激活的核心兴趣。 [10] 为了识别认证用户在每个行为过程中的异常登录，我们考虑了用户行为序列的最大重合部分和短重合。 [11] 基于用户行为序列的推荐变得越来越普遍。 [12] ，识别有影响力的个体并理解用户行为序列之间的交互。 [13] 当涉及到长序列的用户行为数据时，事情变得更加困难，因为系统延迟和存储成本随着用户行为序列的长度近似线性增加。 [14] 经过两个十年的高速发展，如今成熟的互联网服务平台上用户行为序列的累积已经从用户首次注册开始变得极长。 [15]
Anonymou Behavior Sequence 匿名行为序列
Session-based recommendation aims to predict next item based on users’ anonymous behavior sequence within a short time. [1] The aim of session-based recommendation is to predict the next-clicked item based on the anonymous behavior sequence. [2] Session-based recommendation is the task of predicting the next item to recommend when the only available information consists of anonymous behavior sequences. [3]基于会话的推荐旨在根据用户在短时间内的匿名行为序列预测下一个项目。 [1] 基于会话的推荐的目的是基于匿名行为序列来预测下一个点击的项目。 [2] 基于会话的推荐是在唯一可用的信息由匿名行为序列组成时预测下一个要推荐的项目的任务。 [3]
Historical Behavior Sequence 历史行为序列
However, existing works leverage the social relationship to aggregate user features from friends’ historical behavior sequences in a user-level indirect paradigm. [1] Using this multi-dimensional vector representation, items related to a user historical behavior sequence can be easily predicted. [2] Sequential recommendation system's goal is to predict users' next actions based on their historical behavior sequences. [3]然而，现有的作品利用社会关系在用户级间接范式中从朋友的历史行为序列中聚合用户特征。 [1] 使用这种多维向量表示，可以轻松预测与用户历史行为序列相关的项目。 [2] 顺序推荐系统的目标是根据用户的历史行为序列来预测用户的下一步动作。 [3]
Heterogeneou Behavior Sequence 异质行为序列
We propose first integrating the user’s behaviors in search and recommendation into a heterogeneous behavior sequence, then utilizing a joint model for handling both tasks based on the unified sequence. [1] The proposed LSTM is able to learn the student profile-aware representation from heterogeneous behavior sequences. [2]我们建议首先将用户在搜索和推荐中的行为整合到一个异构的行为序列中，然后利用一个基于统一序列的联合模型来处理这两个任务。 [1] 所提出的 LSTM 能够从异构行为序列中学习学生档案感知表示。 [2]
behavior sequence analysi 行为序列分析
The current research, therefore, analyzed observable behaviors leading-up to violent episodes, and used Behavior Sequence Analysis to highlight the typical chains of behaviors that tend toward violence. [1] The current research investigates the life histories of adolescents who have committed murder, using two leading temporal methods: Behavior Sequence Analysis and Crime Script Analysis. [2]因此，当前的研究分析了导致暴力事件的可观察行为，并使用行为序列分析来突出倾向于暴力的典型行为链。 [1] 目前的研究使用两种主要的时间方法来调查犯下谋杀罪的青少年的生活史：行为序列分析和犯罪脚本分析。 [2]