Building an effective recommender system requires more than just a strong model; it involves addressing a range of complex technical issues that contribute to the overall performance. This chapter explores recommender systems from seven distinct angles, covering feature selection, retrieval layer strategies, real-time performance optimization, scenario-based objective selection, model structure improvements based on user intent, the cold start problem, and the “exploration vs. exploitation” challenge. By understanding these critical aspects, machine learning engineers can develop robust recommender systems with comprehensive capabilities.

Scientific knowledge is abundant, but this abundance has created challenges. What can be synthesized from the research is limited because of the inconsistent use of terms and classification systems. For example in clinical research, literature reviews, such as meta-analyses, are critical in the development of clinical practice guidelines and recommendations. And the problem is especially acute in the behavioral sciences, where the lack of an agreed-upon classification system for research terms means this knowledge is less likely to be synthesized and interpreted in a manner that can affect clinical care and public policies. This chapter examines the gap between what is known and the capacity to act on that knowledge. We discuss strategies to make research more replicable, better organized, and more easily retrieved.

The formal details of the modeling frameworks that have been most useful in accounting for specific empirical phenomena are presented. At the highest, most abstract level are mathematical models used to describe how the contents of the short-term store are managed. At the middle level, the search of associative memory models (SAM) describe how information is transferred from the short-term store to the long-term store, and how memories in the long-term store are retrieved to the short-term store. At the lowest, most complex level, the retrieving effectively from memory (REM) models are described, which implement multidimensional memory representations and rational decision processes.

The law should be ‘computable’, in order to make retrieval and analysis easier. Computable law takes aspects of law, which are implicit in legal texts, and aims to model them as data, rules, or forms which are amenable to computer processes. Laws should be labelled with computable structural data to permit advanced computational processing and legal analysis.

This brief introduction flags up the problems of song recovery from eras before the advent of music publishing and mechanical recording. It pits assumptions of European colonial superiority against the voices and musical practices of America’s Indigenous people, from the Inuits of Alaska to the Aztecs of Mexico.

The main idea behind this chapter is that the research on deliberately learning vocabulary shows that it is such an efficient and effective way of learning that, in terms of speed of vocabulary growth, a course that includes both incidental learning from communication-focused activities and deliberate learning will be much better than a course that relies only on incidental learning. Moreover, flash-card learning should be done so that the vocabulary worked on is repeatedly met until it is thoroughly learned. This goal of a high level of knowledge contrasts with the uncertain learning that occurs through other activities. Deliberately learning vocabulary before it is met in communication-focused activities means that the difficulty of such activities is reduced. Deliberately learning vocabulary after it has been met in a communication-focused activity means that the deliberate learning will be more meaningful and motivated. This chapter describes a list of well-supported guidelines for learning using flash cards. These principles should also be used in electronic flash-card programs.

The main message of this chapter is that vocabulary teaching has only a limited role to play in the learning of vocabulary (no more than one-quarter of one of the four strands of a course), and teachers need to be careful not to overplay this role at the expense of the time available for the deliberate learning and incidental learning of vocabulary. In this chapter, the role of teaching is very narrowly defined, and it is used to refer to the teacher deliberately teaching vocabulary. The conditions needed for vocabulary learning are carefully examined and two approaches to evaluating vocabulary teaching activities, involvement load and technique feature analysis, are described and applied. The chapter includes a range of practical activities for teaching words, including concordancing and computer-assisted vocabulary learning. These different activities are related to the various aspects of knowing a word.

The ability of Irish poetry to secure a place in the canon is intimately tied to structures of power and authority – structures with a strongly gendered character. Historically, the Irish tradition has been fractured by numerous traumas and the programmatic exclusion of poetry by women. Anthologies are key drivers in canon formation, but have been heavily slanted against equal representation of male and female poets. Often, women will achieve representation in one anthology only to disappear from view in the next; and applying a historical overview, we find that the twentieth century is often worse in this regard than previous eras. A number of case studies are considered, with close attention to questions of group dynamics, literary movements, citationality, and the role of the academy and the anthology.

Memories may spontaneously decay over time, but the main cause of forgetting appears to be interference from other memories. Many apparently lost memories are still in storage, but preceding and following events can both make it harder to find them (proactive and retroactive interference), especially if the memories are similar. Retrieval is more likely insofar as our current environment—including our emotions—resembles the original; stimuli that were present then can act as retrieval cues, reactivating associated memories. If a retrieval cue is associated with many memories, however, it is less likely to activate any one of them (cue overload). While a memory is being activated, it may automatically inhibit competing memories (retrieval-induced forgetting), and we can also suppress memories deliberately. If we are only able to activate fragments of a memory, we may unconsciously reconstruct the original, and research on the misinformation effect, source confusion, and reality monitoring has shown how flawed this reconstruction can be, in some cases leading us to confidently recall events that never occurred. The combination of constructive processes during perception and reconstruction during retrieval mean that our memories are more like rough sketches of events than detailed photographs.

Emotional experiences are often more likely than neutral experiences to be remembered, or to be retrieved richly. In this chapter, we provide an overview of how the effects of emotion arise, emphasizing the effects that operate during the initial experience of the event (encoding), the storage and stabilization of the memory trace for that experience (consolidation), and the accessing of that trace (retrieval). We discuss how these effects of emotion can explain both why emotion enhances many aspects of memory throughout the adult life span and also why there are often age-by-valence interactions in memory, with older adults remembering information more positively than younger adults.