Quickly find photos and videos 24 hours a day with Digital Asset Management

Many companies manage dozens of photos and videos on a daily basis that they use for internal and external communication. It's not always easy to find the content you need right away, often hastily saved to a single device and without enough clues to recover.

For those who use a lot of multimedia content, the need to adopt a method in archiving and organizing files is concrete. It is true, it is difficult to rename the files and describe them adequately but it is far from wasted time. In fact, it is a question of laying the foundations for lasting time savings as well as transforming simple files into digital assets. Thanks to a correct metadata management strategy, photos and videos can be easily found and reused. This is what gives value to multimedia material. What to do with great content if it can't be found? It is as if they did not exist.

How to define the best search algorithm

To enhance photos and videos it is essential that they can be found quickly and reused. There are two main elements to achieve this goal:

• use a centralized platform for digital asset management
• define a search algorithm functional to business needs

The search engine is the beating heart of the Digital Asset Management software and returns the results based on the search algorithm implemented.

With MomaPIX DAM the algorithm is agreed with the customer taking into account the type of archive to be treated (product catalogs, creative, editorial, institutional photos, etc.), the quantity and quality of the metadata associated with the images and the tools that the technology uses. layout. The keywording and the maintenance of a possible thesaurus by the customer are therefore crucial elements in order to obtain a good result.

However, even by carrying out these activities in the best possible way and being able to use the best technology, there is no optimal algorithm for any type of research; it is necessary to find the right combination to reach the best compromise.

Operation criteria of the search engine in the DAM

Basically, a search algorithm takes into account one or a combination of more than one of the following:

• Relevance - the photos in which the search keywords entered by the user are most relevant in the metadata with which the photos are described are shown first. Among the first results are shown the images that, for example, have as their only keyword the one sought by the customer rather than the images that contain more keywords in addition to the one sought. The more accurate the metadata is compiled, the more information will be available for image search and consequently the better the search results will be.
• Date - you can make the photos with a more recent date be shown first than the others or vice versa.

As part of the results obtained with these criteria, it is then possible to leverage other tools to change the ranking assigned in the first instance to the contents by the algorithm and thereby ensure that the order of appearance on the screen is different from the calculated one. at first.

Below is the list of additional elements that may come into play:

• Boolean operators - the logical operators AND, OR, NOT allow you to combine several words in the same search in various ways. With AND you search for images in which the keywords entered by the user must all be contained within the metadata; with OR there must be at least one of the searched words (but also all); NOT excludes images containing the specified keyword.

• Synonyms (wordform): a dictionary of synonyms thanks to which the software is able to expand the search (e.g. by typing happy mood - the search can also be automatically extended to good mood as the two combinations are considered similar). It is up to the customer to provide an adequate thesaurus for the keywords that describe their image heritage.

The synonyms, unlike the equivalents (see below) extend the search bidirectionally. That is, looking for makeup or make up you get the same results.

• Equivalent: Extend searches to equivalent words. That is, if, for example, you search for the word Parent, you can also extend the search to all the images that contain Father, Mother, Dad or Mom in the metadata.

Unlike synonyms, the search extension in this case is not bidirectional. That is, if I search for "Father", the search is not extended to all the photos that also contain the word Parent in the metadata, which would also include mothers.

• Stopwords: words that add no meaning to content, such as articles, prepositions and conjunctions. It may be convenient that these words are not considered by the search engine so that the algorithm is more efficient.

• Stemming: process of reducing a word to its root in order to expand the results of a search (eg. Walk - walking - walked). In practice, the engine can consider or not in the search all the declensions of a word starting from its root, including the masculine / feminine, singular / plural forms.

• Ranking: you can decide to assign a particular score to the images, a quality evaluation, so that the images with a higher evaluation, all other things being equal, are presented before the others or in any case in a higher position compared to the one normally returned by the search engine.

• Proximity ranking: applies the logical boolean operator AND between two words but in showing the results it gives greater importance to the images in which the two words, in addition to being both present in the metadata, are close to each other (eg if you search for " single woman "will show first the images that contain the two consecutive words single and woman in the metadata rather than those that contain both but not close)

• Shuffle: allows you to shuffle the results returned by the search engine before showing them to the user. The use of this algorithm is particularly useful in contexts where the archive contains many sequences of shots belonging to the same service with uniform metadata. By applying the normal search criteria by relevance, the images would be assigned the same ranking and therefore would be shown all consecutively. If you want to prevent this from happening, you can decide to apply the Shuffle algorithm to all searches which mixes them with images from other services before showing them on the screen to the user.

Navigation tree and dynamic filters to facilitate the search

Timely keywording, combined with one or more of the elements listed in the previous paragraph, have a decisive impact on the quality of the result that can be obtained and, consequently, on the quality of the search experience of its users in the archive.

But what else can improve the user experience in research? An important role is played by the definition of the navigation tree (which includes the various levels of categories) and dynamic search filters.

MomaPIX DAM provides for an extensive and multifunctional use of filters. For the individual files uploaded to the archive, beyond keywords and metadata in general, it is possible to define attributes, or special fields that can take on a limited set of values and be shown as dynamic filters that will allow the user to further refine your search. For example, if we are talking about a table catalog, the shape can be defined as an attribute and then shown as a filter within the search results.

Filters are dynamic because, as a source of a search, only the attributes that present results with the relative number of occurrences are shown.