0. about alexandria

Alexandria is an early stage startup focused on information retrieval and natural language processing (NLP) for long-form content. Essentially, the question they ask is this:
how can we use machine learning algorithms to make reading books more "internet-like"?

I liked the question a lot (｡•̀ᴗ-)✧ so I decided to take on the role as sole designer for Alexandria's first product.

The product consists of three core elements:

the algorithms • capable of querying long-form content, using question-answer modeling, and generating recommendations

the corpus • available through partnerships with publishing houses. accessible to users through a subscription service

the interface • a search engine, an adaptive e-reader, shareable annotations

This case study details my process in creating the Alexandria web application. But there's more !! See how I design the Alexandria landing page here.

I. objectives and constraints

making books more internet-like

The motive behind Alexandria was that technology had progressed to fundamentally change the way we read. NLP advancements had made it possible for an updated reading experience, beyond digitization of text.

By parsing its content library with NLP algorithms, Alexandria could search books like searching the internet, uncovering rare knowledge found only in books. Users could ask questions and receive in-depth answers pulled directly from the text. Then based on the content itself, Alexandria's NLP recommendation system could help readers find everything they didn't know they were looking for.

Such functions were already familiar to the web, but there were many facets to the online experience ! My challenge was to understand them, and coalesce the best of these mediums.

creation from consumption

In vein of technological convergence, the future centered on user-generated content— Alexandria intended to grow into a publishing platform. It would be a while until this became a probable reality, but I wanted to construct something foundationally consistent to Alexandria's vision.

I wondered to what extent could UX incite readers to become authors. When deep into content, how could UX create space for baby ideas to surface? What were the bounds of interacting with long-form, independently versus collaboratively, and how might we reach them?

seed phase prototyping ; innovation over perfection

Alexandria was in its seed phase, which meant operating with limited data, and prototyping for more. For now, we only needed to convey a concept; it’d take experimenting to realize the product’s magnetism. This was the time to try many ideas, to get as much feedback as possible before chain-testing iterations.

II. research

The high level goals were to:

contextual mapping : on reading & using the internet

To make the task less chimeric, I deconstructed the experiences of reading a book and using the internet in a series of context maps.

Each strand connected an aspect of the reading / online experience to a context with specific framing objectives.

analyze the pain points particular to reading books / internet usage

identify compatible aspects in both experiences; balancing the weaknesses of one with the strengths of the other

make tangible the nebulous feelings of sinking into a book, getting sucked into an internet rabbit hole ( and igniting creative impulses )

create a resource I’d feel lucky to have any time I wanted a fresh perspective

competitive analysis

Concurrently with context mapping, I analyzed 9 competitors from 3 categories of relevance—for reading books, using the internet, and for specialized media platforms with fun, compelling features.

III. lo-fi • feedback and revision

well perceived / for keeps --

tab mode vs deep mode

productivity however you mean it
view multiple books and searches with tabs, or focus on full screen

chrome-spotify hybrid

interface felt immediately familiar and intuitive to use

generating click-throughs

experimenting w metrics, rankings, occurrences, snippet length
sufficient info for click interest

to be re-examined --

less of a document, more of a canvas

more demand for pen / sketch tools than anticipated
enabling drag & drop interactions, distinguishing from tethered annotations

urge for greater convergence

users recalled writing down questions while reading and forgetting to search later
wanted stack exchange-type integration, upvote / forum systems

usability testing ( some things right, some things wrong )

Participants were asked to voice their thought processes while figuring out how to:

read a book in two viewing modes
annotate the text, add a bookmark
create a new collection
search / filter and assess the results
seek recommendations

overall --

Users found it a lot easier to critique the e-reader, rather than search result interactions; this would require later testing.

These wireframes were practical, but even in greyscale I felt its structure could be more captivating than it was. I wanted our customers to be curious, always.

Aesthetic direction : minimal enough to be non-distractive, influenced by glass architecture, light & shade, rabbit hole dimensions, feminine robot muses, and classical celestial iconography

IV. hi-fi • visual design and prototyping

UNCONSTRAINED TO ONE SHELF

home • browse / annotations / collections

Your content as you left it; easily retrieve anything.

Browse NLP recommendations that are “made for you”. “Discovery” recommendations & “rabbit hole” categories help realize your next obsession.

Organize your books like playlists.
Customize collections for every mood and thesis.

VIEW IN AN ADAPTIVE READER

tab mode • deep mode

Navigate content like switching tabs OR immerse within a single text.

Flexibly annotate, then share with others.
Tether multimedia annotations to text, or freely draw / drag & drop elements.
Import annotations from authors, friends, and other users.
Export annotations; read together across time and space.

MAXIMIZING RABBIT HOLES

text selection menus / search results

Surf books like you surf the internet.
Each word is an opportunity to go deeper.

Select a word and search with one click.
Select a sentence to find similar passages in other works.
‍
Ask questions get answers without ever closing your book; receive rich in-text answers from trained algorithms, and view upvoted community responses.

V. future directions

refining search & filter

search data aggregated from private beta users should inform on actual patterns of interaction
NLP means that one search from you triggers multiple automated related searches
the way people search web pages may not be the same as how they search books
so what in books do people actually search for? how do they articulate their desires, and alter phrasing to narrow results?

The current filtering system assumes searches on everything, and supports refinement by scope, subject, and format. Scope and format selections are available via drop down menu, while subject functions through text field entries. But this may not be the most compatible !

For instance, it’s possible customers prefer a balance of pre-conceived genres and tags to choose from, in addition to custom tagging, wrangled by human intuition or even AI in the future. As private beta customers continue to train the algorithms, we’ll get a better sense for designing experiments to make book search info more cognitively apparent, and expand on the diversity of rabbit holes.

accessibility concerns

Making books more internet-like necessitates equitable access to information and technology. The next iteration must prioritize inclusive reading spaces.

options for higher color contrast ratios, larger and bolder icons, dyslexia-friendly interface fonts, maintaining aesthetic direction for color-blind readers, foreseeing text-to-speech, keyboard shortcuts

scaling up & down

there could be various reasons why we’d launch an unideal product, from technical complexities to minimum viable product strategies
some shareable annotations can be too demanding too early, i.e. forum capabilities, comments, direct responses also imply profile management and notification systems
things i want to explore next: really utilizing the new tab page, building helpful no results pages, what exactly do you see when you click open a book, what stats do people care about, creator resources, long-form annotations, potential gamification

Rica Chan

game design concept • iji

web application • alexandria

landing page • alexandria

about

contact