We study the problem of pliable private information retrieval with side information (PPIR-SI) for the single server case. In PPIR, the messages are partitioned into nonoverlapping classes and stored in a number of noncolluding databases. The user wishes to retrieve any one message from a desired class while revealing no information about the desired class identity to the databases. In PPIR-SI, the user has prior access to some side information in the form of messages from different classes and wishes to retrieve any one new message from a desired class, i.e., the message is not included in the side information set, while revealing no information about the desired class to the databases. We characterize the capacity of (linear) single-server PPIR-SI for the case where the user's side information is unidentified, i.e., the user is oblivious of the identities of its side information messages and the database structure. We term this case PPIR-USI. Surprisingly, we show that having side information, in PPIR-USI, is disadvantageous, in terms of the download rate, compared to PPIR.

翻译：我们研究了单服务器情况下带有边信息的可塑私有信息检索（PPIR-SI）问题。在PPIR中，消息被划分为不相交的类别并存储在多个非共谋数据库中。用户希望从所需类别中检索任意一条消息，同时不向数据库泄露所需类别的身份信息。在PPIR-SI中，用户预先拥有一些来自不同类别的消息作为边信息，并希望从所需类别中检索任意一条新消息（即该消息不包含在边信息集合中），同时不向数据库泄露所需类别的信息。我们刻画了在用户边信息未识别情形下（即用户不知道其边信息消息的身份以及数据库结构）的（线性）单服务器PPIR-SI的容量，并将此情形称为PPIR-USI。令人惊讶的是，我们发现在PPIR-USI中，相比于PPIR，拥有边信息在下载速率方面反而具有劣势。